Donate
Independent, objective, nonpartisan research

R 804AGR

Authors

R 804AGR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3742 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_804AGR.pdf" ["wpmf_size"]=> string(6) "971557" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(164811) "Evaluating Academic Programs in California’s Community Colleges ••• Andrew M. Gill Duane E. Leigh 2004 Library of Congress Cataloging-in-Publication Data Gill, Andrew M. (Andrew Matthew) Evaluating academic programs in California’s community colleges / Andrew M. Gill, Duane E. Leigh p. cm. Includes bibliographical references. ISBN: 1-58213-100-7 1. California Community Colleges. 2. Community colleges— Curricula—California—Evaluation. 3. Community colleges— Curricula—California—Statistics. I. Leigh, Duane E. II. Public Policy Institute of California. III. Title. LB2361.6.C2G55 2004 378.1'543'09794—dc22 2004017225 Copyright © 2004 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. PPIC does not take or support positions on any ballot measure or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword California’s system of higher education is a three-legged stool. One leg of that stool—the California Community College (CCC) system— has the largest number of students. Sprawling over the state with 108 colleges in 72 districts, the CCC system enrolled well over 1.5 million students in academic year 2002. The other two legs of the stool—the University of California (UC) and California State University (CSU)— together enrolled less than 600,000 students in that same year. PPIC is committed to improving public understanding about the financing and operation of California’s community colleges because they are critically important in educating and training the state’s future labor force. In this study, Andrew Gill and Duane Leigh ask a key question about the function of the CCC system: To what extent do California’s community colleges choose different missions? This is an important question for two reasons. First, as part of the master plan for higher education, it is assumed that community colleges often represent a stepping stone to completion of a four-year education at one of the UC or CSU campuses. Do the community colleges perform this function? If not, alternative means of preparing students for four-year degrees will be required. Second, if community colleges are to be assessed for their contribution to higher education in the state, each campus will have to be judged against its own mission—not some generalized mission that seriously weakens the conclusions of a statewide evaluation of the system. In fact, the authors conclude that a “one-size-fits-all” strategy for community colleges may not be appropriate. Over 40 percent of the colleges specialize in one way or another. For example, Gill and Leigh find that most credits offered by most community colleges are transferable to four-year colleges. They note that both vocational education programs and traditional academic programs generate these credits. At the same time, many campuses provide vocational education iii courses that are not transferable but that are expected to lead to jobs in the local labor market. The authors estimate that 19 of California’s 108 community colleges emphasize nontransferable vocational education credits and de-emphasize transfer credits, whereas 26 community colleges do the opposite. The message of this report is clear: Although community colleges are part of a statewide system of higher education, their roots in local communities generate a greater variety of missions, curricula, and courses than is found in traditional four-year universities. This variety is to be expected, and presumably each campus serves the needs and requirements of its particular student body. It is likely that tensions will continue between the overall goals of the California system of higher education, which is funded in large part with state resources, and the goals of local community colleges, which rightly feel that they are more in touch with the preferences of families and employers in their respective cities and towns. This tension has probably been responsible, in part, for a declining share of total higher education spending allocated to community colleges. The tension is a reflection of those who see community colleges as an integral part of a three-legged stool and of others who prefer that community colleges exercise their freedom to break from the mold that constrains most college curriculum. There is no easy solution, but the authors of this report make it clear that the mission of each community college is important to its program of services and should therefore be judged accordingly. David W. Lyon President and CEO Public Policy Institute of California iv Summary Community colleges have traditionally received funding based on student enrollment, which is usually considered an input in the educational process. Recently, however, legislation enacted at the federal and state levels specifies that funding is to hinge, as least in part, on student performance—an output measure. Performance standards improve resource allocation by identifying colleges whose programs do not measure up to the standards so that remedial action can be taken or sanctions imposed. At the same time, performance standards applied uniformly across community colleges may be counterproductive if colleges differ significantly in their missions. Using data for community colleges in the California Community College System (CCCS), our study provides empirical evidence on the extent to which community colleges choose different missions. We ask three interrelated questions. First, do colleges differ in their missions as reflected in the mix of academic programs they offer? The size and complexity of the CCCS make it likely that individual colleges will choose different missions, which we measure as differences in curriculum “emphasis” and “specialization.” Differences in curriculum emphasis are defined by systematic deviations in the curriculums offered by individual colleges from those offered by a “typical” or average college. Curriculum specialization goes a step further by linking an emphasis on one curriculum measure with a de-emphasis on another. If colleges do appear to differ in their missions, our second question is whether observed differences in emphasis and specialization can be explained using measurable indicators of college-specific characteristics and community needs. If community colleges differ in ways that are linked to these explanatory variables, the third question is whether our evidence may be useful in developing guidelines for evaluating community college performance. v Evidence on the missions of community colleges is typically obtained by interviewing college administrators and stakeholders in the community. The approach we pursue in this report, in contrast, is to collect in one large dataset curriculum information available online for all 108 CCCS campuses. One element of this dataset is our characterization of whether individual colleges have different curriculum emphases as indicated by their campus descriptions. In addition, we collect a wealth of quantitative data on curriculum mix measured in terms of credits, courses, programs, and the interests of freshman students. To this curriculum information, we append data available online measuring local labor market characteristics, community demographic variables, and college-specific characteristics such as membership in a multicampus community college district and proximity to a four-year college. In response to our first research question, we find that most community colleges are oriented toward offering programs that supply credits transferable to four-year colleges. We note, however, that transferable programs include both traditional academic programs and vocational education (voc-ed) programs such as programs in business, computer science, and electronics. Transferable voc-ed offerings are increasingly common in California and nationwide. Where California community colleges do seem to differ is in their emphasis on voc-ed offerings that are transferable and taught at an advanced level. For example, the interquartile range calculated for a curriculum variable measuring the proportion of transferable voc-ed credits indicates that the top 25 percent of colleges differ from the bottom 25 percent by at least 27 percentage points. Similarly, the top 25 percent of colleges differ from the bottom 25 percent by nearly 20 percentage points on the proportion of voc-ed courses taught at an advanced level. Moving beyond emphasis to the concept of specialization, our preferred measure suggests that 45 colleges specialize in some way. Specifically, 26 colleges have a transfer specialization, meaning that they choose to emphasize transfer credits while de-emphasizing nontransferable voc-ed credits. Another 19 colleges specialize in nontransferable voc-ed, meaning that they emphasize nontransferable voc-ed credits while de-emphasizing transfer credits. vi Turning to the second research question, the intercollege differences in curriculum emphasis and specialization we observe are reasonably well explained by measurable indicators of college-specific characteristics and community needs. Colleges that specialize in either a transferable curriculum or in nontransferable voc-ed offerings are more likely than other colleges to belong to a multicampus district and to be located in large labor markets. Between these two categories of specialization, colleges that have a transfer specialization tend to be closer to a UC or a CSU campus. Colleges that specialize in nontransferable voc-ed tend to have a larger proportion of Hispanic students and to be in larger communities with greater minority populations. Building on our empirical evidence, we offer the following observations that we believe should be considered in evaluating the performance of California’s community colleges. 1. On average, California community colleges offer more credits in transferable programs than in other broad curriculum categories. With a few exceptions, our evidence suggests that colleges are heavily engaged in offering transferable curriculums. 2. Although colleges place a priority on transferable offerings, attention to transfer rates is not sufficient for evaluation because transferable curriculums are a mix of voc-ed courses and traditional academic courses. Success for colleges with a strong transferable voc-ed orientation might be placement in trainingrelated jobs as opposed to transfers to four-year colleges. 3. Some colleges specialize in transferable curriculums, whereas others specialize in nontransferable voc-ed. Similarly, colleges differ in their emphasis on advanced voc-ed curriculums and in the interests of their freshman students in voc-ed. These differences in specializations and emphases indicate important differences in the missions chosen by different colleges. Differences in missions, in turn, suggest that a “one size fits all” evaluation strategy is not appropriate. 4. Basic skills programs are a small proportion of most community colleges’ total credits offered. vii 5. Differences in missions as reflected in differences in curriculum specializations and emphases are responsive to differences in college-specific characteristics and community needs. Precisely how to allow for differences in missions in evaluating performance is a difficult issue that is currently receiving attention by policymakers in states across the nation. We conclude our report by describing two possible approaches. The first is a model-based approach that allows common performance standards to be adjusted in response to quantitative differences between colleges in student characteristics and local economic conditions. To the extent that colleges can demonstrate that their missions differ, the second approach is for the state to tailor performance standards to take into account alternative missions. viii Contents Foreword....................................... iii Summary........................................v Figures ....................................... xiii Tables ......................................... xv Acknowledgments................................ xvii 1. INTRODUCTION .............................1 Multiple Missions and Reporting Requirements..........3 Research Objectives .............................6 Organization of the Report ........................7 2. MEASURES OF COMMUNITY COLLEGE CURRICULUM MIX ...........................9 CPEC Campus Descriptions ......................10 PFE Data on Enrollment.........................12 Goal 3 Measures of Broad Curriculum Categories......13 Goal 4 Variables Measuring Categories of Voc-Ed Credits ..................................15 Breaking Out Traditional Academic Transfer Credits ..................................16 VTEA Data Measuring Courses and Programs and Program Inventory Data .....................16 Voc-Ed Course Measures .......................17 Voc-Ed Program Measures ......................18 First-Time Freshmen Data........................19 Summary....................................22 3. CURRICULUM SPECIALIZATION ...............23 Getting Started with Histograms ...................24 What Do IQ Ranges Tell Us About Differences in Emphasis?................................25 Evidence of Specialization ........................29 ix Correlation Result 1: Transferable Curriculum vs. Nontransferable Voc-Ed......................29 Correlation Result 2: Transferable Curriculum vs. Proportion of Voc-Ed That Is Nontransferable ......35 Correlation Result 3: Positive Relationships Between Measures of Advanced Voc-Ed Curriculum ........36 Correlation Result 4: Tradeoff Between Student Interests in Voc-Ed and Transferring.............36 Bringing It Together Using Factor Analysis ............37 Summary....................................40 4. EXPLAINING DIFFERENCES IN CURRICULUM MIX .......................................41 Description of Explanatory Variables ................41 College-Specific Characteristics...................41 Local Job Opportunities........................44 Community Demographics .....................45 Explaining Differences in Specializations and Emphases ................................46 Specializations in Transferable Curriculums and Nontransferable Voc-Ed......................46 Emphasis on Advanced Voc-Ed ..................49 Student Interest in Voc-Ed......................51 Summary....................................52 5. POLICY DISCUSSION .........................53 Key Provisions of the California Master Plan ...........53 Performance Evaluation Requirements ...............54 Guidance to Policymakers for Evaluating Community Colleges .................................56 What does Our Evidence Show? ..................57 Allowing for Differences in Missions ...............60 Appendix A. Technical Detail on Measures of Curriculum Mix .......63 B. Technical Detail on Factor Analysis .................67 C. Data Sources and Technical Details Used to Construct the Explanatory Variables ...........................71 D. Dataset for California Community Colleges ...........75 x References ......................................79 About the Authors ................................81 Related PPIC Publications...........................83 xi Figures 3.1. Ratio of Transferable Credits to All Credits .........24 3.2. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits ............................25 3.3. Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Credits ..................................31 3.4. Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits ............................36 xiii Tables 1.1. Percentage Completion and Transfer Rates for Full-Time Freshmen Students Enrolled in California Community Colleges, Fall 1999 to Spring 2002.......4 2.1. California Community Colleges’ Emphases Based on CPEC Campus Descriptions ...................12 2.2. Measures of Curriculum Mix Using PFE Enrollment Data, 2000–2001 ...........................14 2.3. Measures of Curriculum Mix Using VTEA Course and Program Data and Program Inventory Data, 1998–1999 ...............................17 2.4. Measures of Curriculum Mix Using FTF Student Academic Objectives, 1997 ....................20 3.1. Variability in Curriculum Mix Measures Using Data for Credits, Courses, Programs, and Student Objectives ................................26 3.2. Estimated Correlation Coefficients Between Selected Curriculum Mix Measures.....................30 3.3. Colleges Emphasizing and Specializing in a Transferable Curriculum, Alternative Definitions of Specialization ............................33 3.4. Colleges Emphasizing and Specializing in Nontransferable Voc-Ed, Alternative Definitions of Specialization ............................34 3.5. Top Five and Bottom Five Colleges Ordered by Factor Scores for Factors 1–4 ...................39 4.1. Descriptive Statistics for Explanatory Variables ......42 4.2. Cross-Tabulations Between Specializations in Transferable Curriculums and Nontransferable Voc-Ed and Selected Explanatory Variables .........47 4.3. Cross-Tabulations Between Emphasis on Advanced Voc-Ed Courses and Student Interest in Voc-Ed and Selected Explanatory Variables ..................50 xv B.1. Factor Loadings from the Orthogonal Rotated Factor Patterns .............................68 B.2. Descriptive Statistics for Factor Scores Obtained from Factor Analysis .........................70 D.1. Data Descriptions, Sources, and Descriptive Statistics .................................75 D.2. Construction of Curriculum Mix Variables from Raw Data.................................78 xvi Acknowledgments The financial support of the Public Policy Institute of California is gratefully acknowledged. Lorana Chaparro, Stephanie Lomibas, and Juan Mansilla provided outstanding research assistance. Helpful comments on the first draft of this report were received from Norton Grubb, Christopher Jepsen, Paul Lewis, David Mertes, Fred Silva, and Jon Sonstelie. Joyce Peterson and Peter Richardson provided excellent editorial assistance in rewriting subsequent drafts. Ron Mittelhammer got us off to a good start on our application of factor analysis. We are also grateful for the helpful suggestions of participants at a PPIC seminar held in October 2003. xvii 1. Introduction Community colleges are complex institutions offering a wide range of academic programs to various groups. Historically, community college students completed two years of a general undergraduate education and earned an associate’s degree (A.A.). Students with the interest and capability would then transfer to a four-year college to complete a bachelor’s degree. Kane and Rouse (1999) describe how over time community colleges have also expanded their course offerings in vocational education (voc-ed) and adult basic skills. Voc-ed programs often culminate in an A.A. degree or certificate. But Grubb (1996, p. 94) emphasizes that today, credits earned in such voc-ed programs as business, computer science, and electronics are frequently transferable to four-year colleges. In addition to offering a broad range of courses in traditional academic areas, voc-ed, and adult basic skills, a growing number of community colleges provide “contract training”—that is, occupational skills classes designed to meet the needs of particular employers and sometimes delivered on site. As academic offerings have expanded, so has the number of groups to which community colleges must be responsive. Traditional students are 18–22-year-olds who attend community college full-time with the objective of transferring to a four-year college. Nontraditional community college students include adult workers returning to school to sharpen their skills or earn a degree, dislocated workers and returning homemakers seeking retraining for new careers, single mothers seeking to make the welfare-to-work transition, and immigrants interested in improving their language skills and learning more about American culture. These nontraditional students are typically enrolled part time. Other community college constituencies include employers looking for contract training programs and local government officials seeking to encourage the economic development of their communities. 1 As their program offerings and constituencies expand, community colleges are increasingly being held to performance standards. The primary objective of performance standards—improving resource allocation—is laudable. These standards help colleges identify weak programs and either improve or drop them from the curriculum. They also help community college districts and state systems, such as the California Community College System (CCCS), identify underperforming colleges in need of remedial action or sanctions. The downside of performance standards is that community colleges may not be all alike. When colleges differ substantially in their missions, performance standards applied uniformly will penalize colleges that are accomplishing the mission they have chosen. For example, transfer rate is not an appropriate performance criterion for a community college whose primary mission is to provide occupational training for immediate employment in the local labor market. In this report, we investigate the extent to which California community colleges differ in their missions, which we measure by differences in the mix of academic programs they offer. We focus on California community colleges for four reasons. First, the CCCS is the nation’s oldest and most extensive community college system offering educational services on 108 campuses to over 1.6 million students as of Fall 2001. Romano (2003) remarks that California so dominates national community college statistics that researchers often do their calculations with and without California data included. In a community college system as large and complex as California’s, community colleges are most likely to differ in their missions.1 Second, California is well known for and has been widely emulated by other states for its open admission and low tuition policies. These policies were established in 1960 by the Master Plan for Higher Education in California. Third, the rich ethnic and racial diversity of the state allows us to address issues involving the effect of race and ethnicity on the mix of community _____________ 1As Bailey (2002, p. 69) notes, community colleges cannot be expected to do everything well and therefore must choose a more limited set of objectives in much the same way that a business firm chooses to focus on its “core competency.” At the same time, he indicates that, largely for political reasons, community colleges nationwide often pursue a comprehensive strategy emphasizing breadth of program offerings. 2 college academic offerings. Finally, a wealth of data is available online identifying the academic programs provided by California community colleges. To our knowledge, our report is the first to bring together the several sources of curriculum data for California community colleges into a unified and consistent dataset. Multiple Missions and Reporting Requirements The Master Plan of 1960 gives California community colleges considerable discretion in determining the mix of academic programs to offer. At the same time, California community colleges, as is true for community colleges across the nation, face a plethora of performance standards. One example is the federal student-right-to-know (SRTK) regulation that requires the collection and reporting of transfer and completion rates. To set the stage for our discussion of differences in missions among colleges, Table 1.1 displays transfer and completion rates for 107 colleges in the CCCS measured as of 1999 (Chancellor’s Office, 2003). Several measurement issues make us cautious in comparing the performance of particular colleges using SRTK data.2 The main point we draw from Table 1.1 is simply the variability across colleges in the two rates shown. Completion rates range from a high of 49.5 percent for Las Positas College to a low of 11.5 percent for Los Angeles Mission College. Similarly, transfer rates range from a high of 55.9 percent for Merritt College to a low of 4.6 percent for Lassen College. If we take these data at face value, a question that comes to mind is, what are colleges with low completion and transfer rates doing? One answer is that they are not doing their job and need to improve. An equally plausible answer, and one that is noted in the Chancellor’s Office (2003) report, is that colleges differ in their missions. As suggested above, some colleges may choose not to emphasize the transfer function but instead _____________ 2The first of these is the restriction of the universe of students to first-time, fulltime freshmen (which greatly reduces the tracking pool). The second is a counting methodology that requires a hierarchy of outcomes such that students who earned degrees or became transfer-prepared are removed from the calculation of transfer rates. The third issue is inclusion in transfer rates of students who transfer between community colleges in addition to students who transfer to four-year institutions. 3 Table 1.1 Percentage Completion and Transfer Rates for Full-Time Freshman Students Enrolled in California Community Colleges, Fall 1999 to Spring 2002 4 College Alameda Allan Hancock American River Antelope Valley Bakersfield Barstow Butte Cabrillo Canada Canyons Cerritos Cerro Coso Chabot Chaffey Citrus Coastline Columbia Compton Contra Costa Cosumnes River Crafton Hills Cuesta Cuyamaca Cypress DeAnza Desert Diablo Valley East Los Angeles El Camino Completion Rate 33.7 41.9 33.6 43.3 34.6 29.6 36.8 35.1 23.1 40.7 27.1 39.8 33.8 25.7 28.3 26.7 32.1 17.4 32.5 27.6 29.9 35.3 28.1 20.8 42.8 48.4 41.1 26.1 38.6 Transfer Rate 38.4 11.0 13.4 10.1 23.3 12.7 17.2 17.1 41.0 23.5 20.8 18.3 30.5 18.9 26.8 33.3 30.4 21.7 22.5 25.5 20.5 25.7 28.1 41.2 33.4 4.8 24.3 22.5 16.1 College Evergreen Valley Feather River Foothill Fresno City Fullerton Gavilan Glendale Golden West Grossmont Hartnell Imperial Valley Irvine Valley Lake Tahoe Laney Las Positas Lassen Long Beach Los Angeles City Los Angeles Harbor Los Angeles Mission Los Angeles Pierce Los Angeles Southwest Los Angeles Trade-Tech Los Angeles Valley Los Medanos Marin Mendocino Merced Merritt Completion Rate 28.4 48.5 39.2 30.2 23.7 37.9 32.8 35.5 30.2 44.1 46.4 39.7 35.0 28.0 49.5 23.1 30.2 29.4 29.8 11.5 36.7 23.1 21.6 38.3 24.3 26.5 34.4 30.8 23.7 Transfer Rate 40.1 33.3 31.5 32.8 47.7 22.7 25.5 29.3 24.2 15.3 16.6 23.7 52.5 37.9 22.4 4.6 18.6 25.2 22.8 26.9 30.7 15.4 35.1 28.3 29.2 36.7 23.2 15.2 55.9 Table 1.1 (continued) 5 College Mira Costa Mission Modesto Monterey Moorpark Mt. San Antonio Mt. San Jacinto Napa Valley Ohlone Orange Coast Oxnard Palo Verde Palomar Pasadena City Porterville Redwoods Reedly Rio Hondo Riverside Sacramento City Saddleback San Bernardino San Diego City San Diego Mesa San Diego Miramar Completion Rate 27.8 32.0 34.6 41.5 29.9 31.6 29.8 35.0 33.8 40.3 27.4 31.8 28.2 39.7 41.6 31.7 41.0 23.7 32.1 33.1 38.2 15.7 13.1 28.4 17.1 Transfer Rate 25.9 33.3 18.0 18.7 27.3 28.5 20.2 10.0 26.8 26.7 19.2 18.2 21.0 20.2 13.6 23.2 38.5 32.2 27.2 23.9 22.1 20.4 21.3 30.5 51.4 College San Francisco City San Joaquin Delta San Jose City San Mateo Santa Ana Santa Barbara Santa Monica Santa Rosa Santiago Canyon Sequoias Shasta Sierra Siskiyous Skyline Solano Southwestern Taft Ventura Victor Valley Vista West Hills West Los Angeles West Valley Yuba Completion Rate 43.5 32.4 17.9 41.7 36.3 43.8 36.8 45.7 31.1 34.2 34.7 36.0 30.3 31.9 29.6 36.5 42.0 32.8 23.4 22.1 23.3 17.4 40.6 38.1 Transfer Rate 20.1 17.3 38.4 33.1 25.0 26.6 19.1 9.7 37.8 15.3 16.7 22.0 25.8 24.8 24.4 11.6 14.5 26.6 22.3 48.8 17.0 32.6 31.5 25.3 SOURCE: Chancellor’s Office (2003). NOTES: Completion rates measure the percentage of students who earn an A.A. degree or certificate or become “transfer-prepared.” Excluding those measured as “completers,” transfer rates measure the percentage of students who transfer to a four-year institution or to another two-year college. Copper Mountain College is omitted, since it did not begin operation until 2001. choose to emphasize voc-ed training that provides occupational skills immediately salable in the local labor market. Before leaving Table 1.1, it is worth noting that the dataset used to calculate these rates—the FirstTime Freshman (FTF) student cohort—is discussed in Chapter 2 as a measure of the educational objectives of freshman students. In addition to SRTK regulation, community colleges are subject to reporting requirements imposed by two important pieces of federal legislation. These are the Workforce Investment Act (WIA) of 1998 and the Carl D. Perkins Vocational and Technical Education Act (VTEA) of 1998. These laws impose requirements that are substantially more detailed than those needed to satisfy colleges’ SRTK obligations. In particular, both laws require that states establish comprehensive accountability systems that include multiple “core indicators” of performance. In California, the Partnership for Excellence (PFE) agreement of 1996 between the state and the CCCS and the California Perkins State Plan of 1994 specify performance standards that satisfy federal requirements. With some justification, community college administrators often regard these performance standards as flawed. For our purposes, however, a favorable but unintended consequence of performance standards is that California community colleges report not only outcome measures, such as transfer and completion rates, but also measures of the academic programs they provide. As will be discussed in more detail in Chapter 2, these curriculum measures are expressed in terms of distributions of credits, courses, programs, and student interests. Existing analyses of the question “What do community colleges do?” typically proceed with site visits to selected colleges. Our investigation of a dataset covering all the community colleges in a large system such as the CCCS should allow us to provide a more comprehensive answer to this fundamental question. Research Objectives Our analysis has three primary goals. The first is to measure the extent to which community colleges differ in their missions as measured by the curriculum mixes they offer. We discuss this variation in terms of differences in curriculum “emphasis” and curriculum “specialization.” In 6 Chapter 3, we consider alternative definitions of these two concepts. By differences in curriculum emphasis, we mean systematic deviations in the curriculums offered by individual colleges from that offered by a typical or average college. By differences in curriculum specialization, we go a step further by coupling emphasis on one curriculum variable (for example, traditional academic programs leading to transfer to a four-year college) with a de-emphasis on another variable, such as occupational skills training immediately applicable in the local labor market. The second goal is to explain observed differences in emphasis and specialization using measures of college-specific variables, such as proximity to the nearest four-year college, and of community needs, such as local job opportunities and demographic characteristics of the local service area. Our final goal is to use this evidence to help guide policymakers in evaluating community colleges. Organization of the Report Chapter 2 outlines the several sources of online information we put together into a consistent dataset. This dataset includes, for each of the 108 campuses in the CCCS, a number of alternative curriculum measures. These measures receive detailed examination in Chapter 3, where we look for differences in curriculum emphasis and specialization across colleges. In Chapter 4, we present evidence on the extent to which differences in curriculum emphases and specializations can be accounted for by a reasonable set of explanatory variables including measures of community demographics, local labor market characteristics, and college-specific characteristics. In Chapter 5, we offer an overview of key provisions of the California Master Plan and provide some detail on the multiple performance indicators faced by community colleges. Then we draw on our empirical findings to offer suggestions to policymakers intended to better inform the process of evaluating the performance of community colleges. 7 2. Measures of Community College Curriculum Mix In Chapter 1, we identified the three primary missions of colleges in the California Community College System. This chapter is concerned with measuring these three academic missions using data available online for the 108 CCCS campuses. In this chapter, we present simple descriptive statistics for the curriculum measures developed. A more detailed analysis of these data looking for evidence of curriculum emphasis and specialization is in Chapter 3. One approach to measuring a community college’s primary mission or missions is to examine how the college describes itself on its web page. We begin this chapter with a consideration of these campus descriptions, which are collected on the California Postsecondary Education Commission (CPEC) web site. A second approach to measuring differences in missions is to use online sources of quantitative data measuring curriculum mix. As noted in Chapter 1, California community colleges report a wealth of curriculum data in response to performance standards imposed at the state and federal levels. In this chapter, we describe three sources of quantitative data from which we construct 21 curriculum mix variables. Sources of quantitative data are • PFE data on student enrollment measured in credits; • Data on voc-ed courses and programs reported by the CCCS Chancellor’s Office to satisfy its VTEA obligation; and • FTF data on students’ academic objectives. Added to our discussion of VTEA data is a brief description of the Chancellor’s Office “Inventory of Approved and Projected Programs,” which provides an additional measure of “approved” voc-ed programs. Technical detail on these quantitative datasets is found in Appendix A. 9 CPEC Campus Descriptions For each community college, the CPEC web site provides a brief campus description, information on total enrollment, the gender and race or ethnicity of students, a list of programs offered, and links to additional information about the college. Also reported is a campus mailing address. The campus descriptions provide information indicating what the college believes to be especially noteworthy about its academic programs. Using this information to measure curriculum emphasis, we construct a three-category discrete variable with categories indicating voc-ed emphasis, transfer emphasis, and emphasis indeterminate. The criteria used to classify each college in this threecategory scheme are as follows. Voc-ed emphasis • Pride expressed in particular occupational skill training programs; • Emphasis on partnerships with local business firms; and • Emphasis on employment opportunities in the local area and job placement assistance services. Transfer emphasis • Expression of pride in college’s success in transferring students to University of California (UC) and California State University (CSU) campuses; and • Emphasis on special transfer arrangements with four-year colleges, especially UC and CSU campuses. Indeterminate • Emphasis on breadth of course offerings, student services, and extracurricular programs; • Emphasis on location and facilities; and • Emphasis on other attributes such as multicultural student body, small classes, individual attention, and high-quality faculty. 10 Two colleges that are easily classified as having a voc-ed or transfer emphasis are, respectively, Los Angeles Trade-Tech College and Santa Barbara City College (SBCC). Los Angeles Trade-Tech describes itself as “the ‘flagship vocational college’ of the Los Angeles Community College District.” On the other hand, SBCC writes that “the college prides itself on the overall excellence of its academic programs and student support services. It is a top community college for transferring students to the University of California system.” The SBCC statement goes on to describe the college’s special transfer programs with area four-year colleges and universities. Laden (1999) identifies SBCC and Santa Monica College as the state’s leading community colleges in terms of overall transfer rates and transfers to the highly selective UC system. In the discussion of PFE, VTEA, and FTF data that follows, we use Los Angeles Trade-Tech and Santa Barbara City College as examples to illustrate the variety of curriculum mix measures provided by these datasets. Although Los Angeles Trade-Tech and SBCC neatly fit the criteria outlined above, many other colleges provide campus descriptions that are either too brief or too general to allow ready classification. In these cases, we followed up by examining the college’s mission statement, the president’s introductory letter if available, and other pertinent information. This additional information was seldom helpful. As seen in Table 2.1, we are able to classify 17 colleges, including Los Angeles Trade-Tech, as having a voc-ed emphasis, and 11 colleges, including Santa Barbara City College, as having a transfer emphasis. The remaining 80 colleges fall into the indeterminate emphasis category. Our general impression from reading CPEC descriptions is that most colleges do not take advantage of this opportunity to differentiate themselves when describing their academic programs. Rather, they indicate that they provide a full range of educational programs and attempt to distinguish themselves by emphasizing their location and facilities. 11 Table 2.1 California Community Colleges’ Emphases Based on CPEC Campus Descriptions College with Transfer Emphasis Chabot Diablo Valley Fullerton Las Positas Los Angeles Harbor Los Angeles Pierce Marin Oxnard Santa Barbara City Santa Monica Ventura College with Voc-Ed Emphasis Coastline Contra Costa Copper Mountain Cosumnes River Desert East Los Angeles Foothill Fresno City Lassen Long Beach City Los Angeles Trade-Tech Moorpark San Diego Miramar San Joaquin Delta Shasta Siskiyous Taft PFE Data on Enrollment The Partnership for Excellence is an agreement between the state and the CCCS committing community colleges to five quantitative performance goals, two of which are interesting to us. Goal 3 is the system’s commitment to increase overall rates of successful course completion, where courses are broken down by transferable, voc-ed, and basic skills credits. Goal 4 involves a CCCS commitment to enhance statewide workforce development by increasing successful course completion among voc-ed courses classified as apprenticeship, advanced occupational, and introductory occupational. Advanced occupational courses must have an introductory prerequisite in the same program area. Examples of advanced voc-ed courses are dental pathology, legal secretarial procedures, contact lens laboratory, fire hydraulics, and real estate finance. Introductory occupational courses are intended to provide students with entry-level job skills. Some examples of 12 introductory voc-ed courses are principles of advertising, clinical techniques, and technical engineering. Goal 3 Measures of Broad Curriculum Categories PFE data provided in response to Goal 3 are especially valuable because these data measure the quantitative importance of all three primary community college missions. Transferable courses are defined as courses that are transferable to UC or CSU campuses. Voc-ed courses, in contrast, are defined as nontransferable courses implying a “narrow” definition of voc-ed. Basic skills courses include those defined as “possibly occupational” and “nonvocational.” It should be emphasized that the transfer, narrow voc-ed, and basic skills categories are mutually exclusive. We use credits for courses attempted (as opposed to credits for courses completed) as our PFE curriculum mix measure because credits for courses attempted represent the academic services a college offers. One further point to note about Goal 3 PFE data is that the sum of transferable, narrow voc-ed, and basic skills credits attempted does not sum up to total credits attempted. We define as a “residual” category of course enrollment the difference between total credits attempted and the sum of enrollment in transferable, narrow voc-ed, and basic skills courses. Residual courses seem to us to fall under the heading of “consumption goods,” defined as courses in which students choose to enroll to enhance the quality of their lives as opposed to improving their labor market opportunities or building up transferable credits. The first four rows of Table 2.2 display means and maximum and minimum values for ratios of transferable, narrow voc-ed, basic skills, and residual credits to all credits. Also shown are the ratios calculated for SBCC and Los Angeles Trade-Tech. Row 1 indicates that on average, about 73 percent of credits offered during the 2000–2001 academic year are transferable. The ratio calculated for SBCC lies somewhat above this mean, whereas the ratio for Los Angeles Trade-Tech falls somewhat below. A maximum value of 92.2 percent is found for Orange Coast College in Orange County, and the minimum of 18.7 percent is calculated for Taft College at the southern end of the San Joaquin Valley. It is interesting to note from Table 2.1 that we failed to classify Orange Coast College as having a transfer emphasis because of its brief 13 Table 2.2 Measures of Curriculum Mix Using PFE Enrollment Data, 2000–2001 Enrollment Measure 1. Transferable credits/all credits 2. Narrow voc-ed credits/all credits 3. Basic skills credits/all credits 4. Residual credits/all credits 5. Broad voc-ed credits/ all credits 6. Narrow voc-ed credits/ all voc-ed credits 7. Apprenticeship credits/ all voc-ed credits 8. Advanced voc-ed credits/ all voc-ed credits 9. Academic transfer credits/ all credits SBCC .854 Los Angeles TradeTech .624 Mean .733 Minimum Maximum .187 .922 .058 .162 .087 .000 .756 .041 .093 .069 .008 .215 .047 .121 .111 .015 .326 .254 .252 .229 .063 .821 .227 .643 .356 .000 .920 .000 .057 .017 .000 .727 .614 .180 .207 .000 .647 .658 .534 .591 .117 .859 and nonspecific CPEC campus description. However, we did classify Taft College as having a voc-ed emphasis on the basis of its CPEC description. The remaining three categories of credits shown in rows 2–4 are, on average, roughly evenly distributed across the narrow voc-ed, basic skills, and residual categories. In each of these three rows, in addition, ratios calculated for Los Angeles Trade-Tech are larger than those for SBCC. A large range of 75.6 percentage points between maximum and minimum values appears in row 2 for narrow voc-ed. The maximum value for this ratio is calculated for Taft College, and the minimum value is for Orange Coast College. Maximum/minimum ranges are much smaller for basic skills credits (20.7 percentage points) and residual credits (31.1 percentage points). The mean for proportion of basic skills credits (6.9%) in Table 2.2 probably understates the resources that 14 colleges devote to basic skills, since basic skills programs are typically noncredit programs (Gumport, 2003). Goal 4 Variables Measuring Categories of Voc-Ed Credits Whereas Goal 3 PFE data offer a more complete view of curriculum mix, Goal 4 PFE data provide a detailed breakdown of voc-ed credits by transferability and type. As indicated in Chapter 1, Grubb (1996) emphasizes that a growing proportion of voc-ed enrollment is in courses that are transferable to four-year colleges. He notes, in particular, that courses in business, computer science, and electronics are frequently transferable. A comparison of rows 2 and 5 of Table 2.2 indicates that, measured at the means, the proportion of voc-ed credits to all credits nearly triples to 22.9 percent when we consider the broad rather than the narrow definition. In row 5, inclusion of transferable voc-ed makes the broad voc-ed ratios for SBCC and Los Angeles Trade-Tech essentially the same at about 25 percent, whereas row 2 indicates a much lower narrow voc-ed ratio for SBCC. This difference in mix of voc-ed credits is highlighted in row 6 in the ratio of narrow voc-ed credits to all voc-ed credits. Almost two-thirds of voc-ed credits measured for Los Angeles Trade-Tech are nontransferable, as opposed to less than 25 percent for Santa Barbara City College. Rows 7 and 8 make use of Goal 4 PFE information on credits broken down by type of occupational education courses. Row 7 shows that SBCC has no enrollment in apprenticeship courses, whereas only 5.7 percent of total voc-ed enrollment at Los Angeles Trade-Tech is in apprenticeship courses. Well over half of all colleges resemble SBCC in having no students enrolled in apprenticeship programs. Nevertheless, apprenticeship training does represent a sizable component of voc-ed enrollment at a handful of colleges. The leading example is Santiago Canyon College, in Orange in Southern California, in which 72.7 percent of voc-ed enrollment is in apprenticeship courses. Santiago Canyon is clearly an outlier, however, because the next highest ratio of apprenticeship programs is just 12.3 percent for American River College. Row 6 shows that a much larger proportion of voc-ed credits offered at Santa Barbara City College is transferable than is the case at Los Angeles Trade-Tech. Row 8 indicates, similarly, that many more credits 15 are taught at the advanced level at SBCC than at Los Angeles TradeTech. The minimum level of 0 percent observed in row 8 is for Lake Tahoe Community College, followed by Taft College with only 2.8 percent of voc-ed enrollment at the advanced level. The maximum value of 64.7 percent is observed for Rio Hondo College. Breaking Out Traditional Academic Transfer Credits Knowing both nontransferable voc-ed credits and total voc-ed credits, we can readily calculate transferable voc-ed credits, which, when subtracted from total transfer credits, yields a measure of traditional academic transfer credits. Examples of traditional academic transfer courses are introduction to biology and introduction to sociology. The ratio of academic transfer credits to all credits is shown in line 9 of Table 2.2. Measured at the means in rows 1 and 9, academic transfer credits make up on average about 81 percent (.591/.733) of all transferable credits. The ratio of academic transfer credits to all credits calculated for Santa Barbara City College in row 9 exceeds that for Los Angeles TradeTech by 12.4 percentage points. The even larger difference of 23 percentage points observed in row 1 is clearly due to the greater emphasis of SBCC on transferable voc-ed courses. Minimum and maximum values of the academic transfer credit ratio are observed for Los Angeles Mission College (11.7 percent) and Orange Coast College (85.9 percent), respectively. VTEA Data Measuring Courses and Programs and Program Inventory Data We view the PFE enrollment data summarized in Table 2.2 as approximating an “equilibrium” description of a college’s curriculum mix, where the term equilibrium is used to mean a balance between current student demand and the college’s supply of courses and programs. Recognizing that current offerings of courses and programs hinge in part on past student demand, we view VTEA data as primarily capturing the current supply of courses and programs. From this perspective, VTEA data give us an alternative and independent measure of voc-ed curriculums. 16 Voc-Ed Course Measures Beginning with VTEA course data, rows 1 and 2 of Table 2.3 display ratios of apprenticeship and advanced occupational courses to total voc-ed courses, respectively. The maximum value for the apprenticeship course ratio in row 1 is obtained for Santiago Canyon College. Recall from the discussion of Table 2.2 that Santiago Canyon was singled out as an outlier for our measure of apprenticeship credits to total voc-ed credits. Santa Barbara City College is seen in the second row to have a much higher ratio of advanced occupational courses to all voc-ed courses than is the case for Los Angeles Trade-Tech. The maximum value of this ratio (83.3 percent) is obtained for Mt. San Jacinto College. Note that the sample size in Table 2.3 is 107 rather than 108 campuses because data for Copper Mountain College are not available. Since VTEA data do not provide information on total courses, we were unable to calculate a ratio of voc-ed courses to all courses Table 2.3 Measures of Curriculum Mix Using VTEA Course and Program Data and Program Inventory Data, 1998–1999 Course or Program Measure 1. Apprenticeship courses/ all voc-ed courses 2. Advanced voc-ed courses/ all voc-ed courses 3. Voc-ed courses/100 students 4. Apprenticeship programs/ 100 students 5. Advanced voc-ed programs/100 students 6. Advanced courses/ advanced programs 7. Approved voc-ed programs/total programs SBCC .004 .628 10.608 .007 .288 23.116 .581 L.A. TradeTech .072 .193 12.150 .038 .218 10.765 .952 Mean .026 .265 4.796 .014 .188 6.128 .680 NOTE: Copper Mountain College is omitted. Minimum Maximum .000 .485 .008 .833 .927 22.060 .000 .101 .034 .788 1.750 38.444 .325 .970 17 comparable to that shown for credits in row 5 of Table 2.2. What we can do is express voc-ed course data on a per student basis, using CPEC Fall 2001 student headcounts. Row 3 of Table 2.3 indicates that SBCC and Los Angeles Trade-Tech each offer roughly 11 to 12 voc-ed courses per 100 students. Ratios of this magnitude look large (the mean is under five courses per 100 students), but the maximum ratio is over 22 courses per 100 students observed for Lake Tahoe Community College. Voc-Ed Program Measures Rows 4 and 5 of Table 2.3 make use of VTEA data measuring programs offered. Because VTEA data do not provide a reliable measure of total voc-ed programs, these rows report, respectively, apprenticeship and advanced occupational programs per 100 students. Row 4 continues the impression yielded by our credit and course data that apprenticeship programs are atypical. Row 5 offers a simple way to understand variation in advanced voc-ed programs, which is to “blow up” the ratios shown for a hypothetical college of 10,000 students. For such a hypothetical college, the extreme values suggest a sizable range of between 3 and 79 advanced programs offered. The minimum value of this ratio is observed for Palo Verde College and the maximum for Lassen College. Row 6 shows the “depth” or “intensity” of voc-ed programs measured as the ratio of advanced courses to advanced programs. For this measure, both Santa Barbara City College and Los Angeles TradeTech provide voc-ed programs of greater than average depth. Compared to the mean of about six advanced courses per advanced program, SBCC offers a ratio of 23 and Los Angeles Trade-Tech a ratio of nearly 11. The maximum value of over 34 advanced courses per advanced program appears for Cuesta College. In addition to VTEA data on voc-ed programs offered, the Chancellor’s Office maintains an “Inventory of Approved and Projected Programs.” The program inventory dataset includes all programs that are approved in the sense that a program must require 18 or more credits of coursework. In row 7 of Table 2.3, we use program inventory data to calculate for “approved” programs the ratio of voc-ed programs to all programs. This ratio for Los Angeles Trade-Tech is an astounding 95.2 18 percent. Even Santa Barbara City College reports that 58.1 percent of its total approved programs are voc-ed. Recall from Table 2.2 that the proportion of voc-ed credits to all credits for both colleges is about 25 percent. It is clear that voc-ed programs tend to be much smaller than other programs in terms of student credit hours. First-Time Freshman Data FTF data are derived from a longitudinal study following the universe of 222,372 first-time freshmen at all CCCS campuses between their initial enrollment in Fall 1997 and Spring 2000. We use FTF cohort data measuring students’ academic objectives formed without benefit of academic counseling to represent the demand of incoming students for community colleges’ academic programs. The data distinguish 14 categories of academic objectives, which we group into the following four broad categories of programs. Voc-ed 1. Formulate career interests 2. Vocational degree, no transfer 3. Acquire job skills 4. Update job skills 5. Vocational certificate 6. Maintain license Transfer 7. Degree and transfer 8. Transfer, no degree Basic skills 9. Basic skills 10. Education development 11. Complete GED Other 12. Degree, no transfer 13. Undecided 14. Uncollected 19 Data measuring academic objectives are available for 107 campuses. Again, Copper Mountain College is excluded. Of the four broad categories of programs distinguished for FTF data, probably the most clearly defined is interest in basic skills. As shown in row 1 of Table 2.4, the mean of the proportion of students with a basic skills objective is about 11 percent, which is reasonably close to the 7 percent shown in Table 2.2 for basic skills credits to all credits. Note that the denominator of this proportion excludes the “uncollected” category listed under “other” programs.1 No interest at all in basic skills is reported for freshmen at three colleges, including Canyons, Cerritos, and Citrus. Canyons College is highly unusual, however, because nearly 95 percent of freshman academic interests are undecided (see the Table 2.4 Measures of Curriculum Mix Using FTF Student Academic Objectives, 1997 Measure of Academic Objectives 1. Interest in basic skills SBCC .142 L.A. TradeTech .083 Mean Minimum Maximum .111 .000 .790 2. Interest in transferring .569 .111 .385 .033 .705 3. Interest in voc-ed .143 .644 .252 .017 .644 4. Interest in nontransfer programs .204 .686 .302 .018 .686 5. Undecided .086 .121 .202 .000 .949 NOTE: Copper Mountain College is omitted. _____________ 1Most campuses report either no data uncollected for their first-time freshmen or uncollected data for only a handful of students. However, FTF data are uncollected for more than 5 percent of first-time freshmen at 20 campuses; of these campuses, the uncollected rate is occasionally quite high. In particular, uncollected rates of 10 percent or higher are reported for 11 colleges, including at the high end Rio Hondo College (30.0%) and Sierra College (34.1%). We have no reason to expect that a systematic bias is causing these colleges to report high rates of uncollected responses. Therefore, we proceed with the proportions presented in Table 2.4 using as the denominator total students minus students for whom data could not be collected. 20 maximum in row 5). Minimum values shown in rows 2–4 are also observed for Canyon College. Taft College records the maximum value of the basic skills ratio (79%). We noted in discussing Table 2.2 that Taft College ranks very high in the proportion of voc-ed credits that are not transferable. An unexpected result in row 1 is that student interest in basic skills is slightly higher for Santa Barbara City College than for Los Angeles Trade-Tech. Row 2 of Table 2.4 reports the ratio of freshman students with an interest in transferring to a four-year institution, either with or without an A.A. degree. Excluding Canyon College, the minimum value for this ratio is 7.3 percent for Napa Valley College followed by Taft College at 8.5 percent. The maximum value of 70.5 percent is found for Citrus College. Transfer ratios calculated for Los Angeles Trade-Tech and Santa Barbara City College are close to the minimum and maximum values, respectively, indicating an even bigger difference between the two colleges in interest in transfer courses (45.8 percentage points) than in proportions of transferable credits actually taken (23.0 percentage points). FTF data allow only a narrow definition of voc-ed, which may well be appropriate because many freshman students are likely to be unaware that voc-ed courses are often transferable. In row 3, the mean voc-ed ratio is about one-quarter, with the maximum value of 64.4 percent appearing for Los Angeles Trade-Tech. Excluding Canyons College, the minimum value is for Taft College at 6 percent. This might seem surprising given what we know about Taft, but recall that nearly 80 percent of Taft freshman expect to enroll in basic skills programs. A huge 50 percentage point difference appears between the interest in voced of Los Angeles Trade-Tech freshman as opposed to Santa Barbara City College freshman, with only 14 percent of SBCC freshman expressing an interest in voc-ed programs. In row 4, we add students interested in an A.A. degree (but not in transferring) to those who express an interest in voc-ed. About 30 percent of all freshman report interest in this broader category of nontransfer programs, with the minimum value (excluding Canyons College) observed for Taft College and the maximum for Los Angeles Trade-Tech. 21 Finally, row 5 indicates that, on average, about 20 percent of freshmen students are undecided about their academic objectives. Undecided ratios calculated for both Santa Barbara City College and Los Angeles Trade-Tech are considerably below this mean. No students in the undecided category are observed for two campuses (Cerritos and Citrus), and the maximum value of 94.9 percent is, as noted, recorded for Canyons College. The next highest ratio is 59.9 percent calculated for San Francisco City College. Note that for the first three columns of Table 2.4, rows 1, 2, 4, and 5 sum to 100 percent, subject to rounding error. Summary This chapter describes our attempt at using CPEC campus descriptions to classify California community colleges by curriculum emphasis. This information allows us to measure curriculum mix by what colleges say about themselves in promotional materials. Our conclusion is that 28 out of 108 colleges indicated either a voc-ed or transfer emphasis. We next considered the wealth of quantitative data available online measuring curriculum in terms of credits, courses, programs, and student interests. In contrast to CPEC campus descriptions, these data allow measurement of what colleges actually offer their students. Descriptive statistics for curriculum mix variables constructed from the quantitative data indicate that, on average, nearly three-quarters of all credits in courses attempted are transferable. Measured at the extremes of the distribution, there appears to be considerable variation about this mean, as is the case for other curriculum measures including emphasis on voced, type and transferability of voc-ed offerings, and apprenticeship training. We also noted sizable differences in the mix of voc-ed offerings and interests of freshmen students between Los Angeles Trade-Tech and Santa Barbara City College—our “representative” colleges with voc-ed and transfer emphases, respectively. Chapter 3 examines whether these differences in curriculum mix are pervasive across colleges or are limited to colleges at the extremes. 22 3. Curriculum Specialization This chapter uses the quantitative data described in the previous chapter to identify colleges that differ in curriculum emphases as well as those that offer different curriculum specializations. We define differences in curriculum emphasis as systematic deviations in the curriculum offered by individual colleges from that of a typical or average college. Differences in curriculum specialization couple emphasis on one curriculum variable with a de-emphasis on another variable. For example, a college would specialize if it emphasizes a transferable curriculum while de-emphasizing nontransferable voc-ed. Note that all colleges that specialize must also have an emphasis, but not all colleges that have an emphasis meet our definition of specialization. In the first of four sections, we begin by describing differences between colleges in their curriculum offerings using histograms. Our purpose is to see whether the large spreads observed for many of our curriculum variables are due to only a small number of colleges lying far from means or whether a more substantial number of colleges deviate systematically from means. In the next section, we attempt to distinguish those curriculum variables for which colleges differ substantially in their emphasis. The tool we use to measure differences in emphasis is the interquartile (IQ) range, which contrasts the top 25 percent of colleges with the bottom 25 percent. In the third section, our definition of specialization is implemented by examining the relationships between selected curriculum measures making use of scatter diagrams and pair-wise correlations. This analysis is carried out for selected curriculum measures. The fourth section highlights results of a factor analysis of all of our curriculum measures intended to provide further evidence on curriculum emphases and specializations. 23 Getting Started with Histograms To get a feel for the importance of intercollege differences in curriculums, it is useful to examine visually the distributions of two of our PFE curriculum measures—transferable credits to all credits and narrow (or nontransferable) voc-ed credits to all voc-ed credits. Figure 3.1 displays a histogram for transferable credits. The figure shows that colleges with proportions of transferable credits of 0.5 or less are highly atypical. Instead, the mass of the distribution lies in the range between 0.6 and 0.9. As indicated in Table 2.2, this variable has a mean of 0.733, with a substantial range between a minimum value of 0.187 and a maximum of 0.922. Figure 3.2 shows the histogram for our measure of nontransferable voc-ed credits to all voc-ed credits. Compared to Figure 3.1, the “spread” of this distribution is much larger, so that more colleges are found near the extremes of the distribution. Figure 3.1 indicates that California community colleges are similar in their emphasis on the transfer mission, but keep in mind that for many colleges, transfer credits include a heavy dose of transferable voc-ed coursework. For only a small number of colleges do we observe substantial deviations from the mean. On the other hand, Figure 3.2 illustrates that substantial deviations from 40 30 Frequency 20 10 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 Ratio Figure 3.1—Ratio of Transferable Credits to All Credits 1.0 24 Frequency 20 15 10 5 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 Ratio Figure 3.2—Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits the mean are much more common, suggesting that colleges differ in their emphasis on nontransferable voc-ed. What Do IQ Ranges Tell Us About Differences in Emphasis? Table 3.1 presents several commonly used measures of variability for our 21 curriculum mix variables, where, as in Figures 3.1 and 3.2, we rely on measures of variability to capture differences in curriculum emphasis across colleges. In the first column, we carry over from Tables 2.2 to 2.4 the maximum/minimum range. Standard deviation is reported next. In the remaining three columns, we add the upper and lower quartiles and the associated interquartile range. As a measure of variability, IQ range has an advantage over the standard deviation of not being sensitive to extreme observations. Based primarily on the IQ range, our main conclusion is that colleges differ in curriculum emphasis primarily in terms of the transferability and advanced level of their voc-ed curriculums. On the other hand, we find that differences in emphasis across colleges are small for measures of apprenticeship training and basic skills and “residual” courses. 25 Table 3.1 Variability in Curriculum Mix Measures Using Data for Credits, Courses, Programs, and Student Objectives Curriculum Measure 1. Nontransferable voc-ed credits/all voc-ed credits 2. Advanced occupational credits/all voc-ed credits 3. Advanced occupational courses/all voc-ed courses 4. Approved voc-ed programs/total approved programs 5. Advanced courses/ advanced programs 6. Voc-ed courses/100 students 7. Advanced occupational programs/100 students 8. Interest in transferring 1. Broadly defined voc-ed credits/all credits 2. Transferable credits/all credits 3. Academic transfer credits/all credits 4. Interest in nontransfer programs 5. Interest in voc-ed 6. Undecided interest 7. Interest in basic skills Max/Min Standard First Range Deviation Quartile A. High Variability .920 .198 .207 .647 .142 .101 .825 .158 .153 .645 .140 .576 36.694 4.576 3.429 21.133 4.576 2.465 .754 .114 .114 .672 .127 .296 B. Moderate Variability .758 .105 .162 .735 .106 .693 .742 .111 .547 .668 .109 .229 .672 .104 .183 .949 .114 .149 .790 .094 .055 Fourth Quartile .481 .283 .348 .771 7.357 5.675 .231 .476 .260 .788 .667 .372 .310 .244 .137 IQ Range .273 .182 .195 .196 3.929 3.210 .116 .180 .098 .095 .120 .126 .095 .082 26 Table 3.1 (continued) Max/Min Standard First Fourth IQ Curriculum Measure Range Deviation Quartile Quartile Range C. Low Variability 1. Nontransferable voc-ed credits/all credits .756 .093 .036 .100 .064 2. Apprenticeship credits/ all voc-ed credits .727 .073 .000 .007 .007 3. Basic skills credits/all creditsa .207 .038 .081 .119 .039 4. Residual credits/all creditsa .311 .055 .076 .139 .063 5. Apprenticeship courses/ all voc-ed courses .485 .061 .000 .019 .019 6. Apprenticeship programs/100 studentsa .101 .021 .000 .021 .021 aIndicates variables classified on the basis of a small maximum/minimum range. To delve into the details of Table 3.1, we began by placing in the low variability category three variables with small maximum/minimum ranges: namely, basic skills credits to all credits, residual credits to all credits, and apprenticeship programs per capita. These variables appear in rows C.3, C.4, and C.6. The remaining 18 measures are candidates for the high or moderate variability categories. Fifteen of these remaining 18 variables are measured as proportions, two are measured in per capita terms, and one is program depth measured as courses per program. Beginning with variables measured as proportions, we showed in Figure 3.2 the histogram representing proportion of nontransferable voc-ed credits to all voc-ed credits. In the first row of Table 3.1, the IQ range for this variable (0.273) is seen to be the largest of the IQ ranges shown for all 15 proportionate variables. Its interpretation is that the top 25 percent of colleges (or 27 colleges) report a proportion of nontransferable voc-ed credits to all voc-ed credits that exceeds the proportion offered by the bottom 25 percent by at least 27.3 percentage points. A total of five variables measured as proportions are 27 classified as having high variability, which we define as IQ ranges between 0.180 and 0.273. These variables include • Nontransferable voc-ed credits to all voc-ed credits; • Advanced occupational credits to all voc-ed credits; • Advanced occupational courses to all voc-ed courses; • Approved voc-ed programs to all approved programs; and • Student interest in transferring to all student interests. Seven more variables measured as proportions are classified as having moderate variability as indicated by IQ ranges between 0.082 and 0.142. Note that these variables include the proportion of transferable credits to all credits, the variable depicted in Figure 3.1. The final three variables expressed as proportions have IQ ranges of 0.064 or below, and we place them in the low variability category. Two of these three variables involve apprenticeship training. It is more difficult to classify the remaining three variables involving courses and programs because they are not measured as proportions that must lie between 0.0 and 1.0. Nevertheless, we suggest that the ratio of advanced courses to advanced programs belongs in the high variability category (row A.5). The reason is that voc-ed programs provided by colleges in the fourth quartile (7.4 advanced courses per advanced program) have at least twice the depth as those offered by colleges in the first quartile (3.4 advanced courses per advanced program). We make a similar argument for the per capita measures of voc-ed courses and advanced voc-ed programs (rows A.6 and A.7, respectively). Row A.6 indicates that fourth quartile colleges offer students at least 5.7 voc-ed courses per 100 students, whereas first quartile colleges offer at most 2.5 voc-ed courses per 100 students. Row A.7 suggests that, blown up to the level of a hypothetical college of 10,000 students, fourth quartile colleges offer students at least 23 advanced voc-ed programs, whereas first quartile colleges supply at most 11 advanced voc-ed programs. For both variables, in other words, colleges in the top quartile offer at least twice the number of voc-ed courses or at least twice the number of advanced voc-ed programs as colleges in the bottom quartile. 28 Evidence of Specialization We turn now to the question of whether differences between colleges in curriculum mix represent specialization or an emphasis on one activity with a de-emphasis on another. To implement this definition, we look for evidence of negative relationships between pairs of curriculum variables as measured by correlation coefficients. Negative correlation coefficients are calculated in the two-variable case if, when one variable is above its mean, the other tends to fall below its mean. Table 3.2 displays estimated correlation coefficients calculated for selected pairs of curriculum mix measures. Our focus is primarily on four sets of correlation coefficients, which we label Correlation Results 1–4. Correlation Results 1 and 2 form the basis of our discussion of specialization in this section. Correlation Results 3 and 4, which are only briefly discussed, play a more prominent role in the factor analysis summarized in the next section. Correlation Result 1: Transferable Curriculum vs. Nontransferable Voc-Ed We find a large negative correlation (-0.837) between transferable credits to all credits and narrow voc-ed credits to all credits. That is, colleges above average in transferable credits tend to be below average in narrow voc-ed credits. And colleges above average in narrow voc-ed credits tend to be below average on their transferable credits ratio. This correlation coefficient is reported in row 2 and column (1) of Table 3.2.1 The negative relationship between the transfer credit ratio and the narrow voc-ed ratio is shown graphically in the scatter diagram appearing _____________ 1Recall from Chapter 2 that the transfer credit ratio and the narrow voc-ed ratio are developed from our most inclusive dataset, which we referred to as Goal 3 PFE data. The four mutually exclusive categories of Goal 3 PFE data are transfer, narrow voc-ed, basic skills, and residual credits. A negative relationship between any two of these variables is expected, since proportions representing mutually exclusive categories must sum to 1.0. Hence, being above average on one variable must be offset by being below average on at least one other variable. Nevertheless, a large negative coefficient is not guaranteed. In particular, the correlation between the transfer credit ratio and the basic skills ratio is only –0.09 and not statistically significant. Among the selected variables considered in Table 3.2, the only statistically significant negative correlation coefficient involving basic skills credits is that with freshmen interest in transfer programs. The estimated coefficient is –0.272. 29 30 Table 3.2 Estimated Correlation Coefficients Between Selected Curriculum Mix Measures Curriculum Measure 1. Transferable credits/all credits 2. Nontransferable voc-ed credits/all credits 3. Academic transfer credits/all credits 4. Nontransferable voc-ed credits/all voc-ed credits 5. Advanced occupational credits/all voc-ed credits 6. Advanced occupational courses/all voc-ed courses 7. Advanced courses/ advanced programs 8. Interest in transferring 9. Interest in voc-ed (1) 1.000 –.837 .781 –.766 .237 .383 .139 .355 –.185 (2) 1.000 –.668 .730 –.191 –.284 –.039 –.260 .076 (3) (4) 1.000 –.369 0.000 .162 –.186 .292 –.309 .069 –.068 .415 –.075 –.202 .195 NOTE: Estimates statistically significant at the 5 percent level are in bold. (5) (6) (7) 1.000 .802 1.000 .598 .635 1.000 .167 .270 .292 –.070 –.287 –.236 (8) 1.000 –.494 (9) 1.000 Deviations of transferable credits to all credits in Figure 3.3, where deviations from the mean of transferable credits are measured vertically and deviations from the mean of nontransferable voc-ed are measured horizontally. The negative relationship between these two ratios is captured in the figure by a concentration of data points in the upper left and lower right quadrants. The two outlying data points appearing in the lower right quadrant are for Palo Verde College and Taft College, with Taft College being farther from the origin.2 Recall from Chapter 2 that Taft College places a heavy emphasis on vocational education, particularly nontransferable voc-ed. It is a moderately small college with 8,033 students. Palo Verde College .4 .2 0 –.2 –.4 –.2 0 .2 .4 .6 Deviations of nontransferable voc-ed credits to all credits Figure 3.3—Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Credits _____________ 2Similar outlying data points in the upper left quadrant are not possible because of the large mean of transferable credits to all credits (0.733) and the small mean of narrow voc-ed credits to all credits (0.087). Extreme values of these variables reported in Table 2.2 imply that the maximum deviation from the mean for the transferable credits ratio is 0.189, and the maximum deviation from the mean for the narrow voc-ed credits ratio is –0.087. 31 is a very small college (2,903 students) in Blythe, in the southeastern California desert.3 Continuing with Figure 3.3, it appears that the bulk of the data points are in an imaginary box centered at the origin and extending right and left and up and down by about 5 percentage points. In an attempt to pin down those colleges that specialize as opposed to those that do not, we define colleges that specialize in offering a transferable curriculum as those for which deviations of the transferable credits ratio exceed 0.05 and deviations of the nontransferable voc-ed credits ratio are less than –0.05 (in other words, data points that lie above and to the left of the upper left corner of the box). Similarly, colleges that specialize in offering a nontransferable voc-ed curriculum are those for which deviations of the nontransferable voc-ed credits ratio exceed 0.05 and deviations of the transferable credits ratio are less than –0.05. The names of colleges that, according to these definitions, specialize in a transferable curriculum and in nontransferable voc-ed are reported, respectively, in Tables 3.3 and 3.4. Beginning with Table 3.3, we list in column (1) the names (in alphabetical order) of the 19 colleges that meet our definition of specializing in a transferable curriculum (termed “Definition 1”). For purposes of comparison, we repeat from Table 2.1 the names of 11 colleges that we classified in Chapter 2 as having a transfer emphasis based on our reading of CPEC campus descriptions. These 11 colleges are listed in column (3). There is some overlap between colleges listed in the two columns, with five colleges appearing in common. But it is worth noting that Santa Barbara City College and Santa Monica College are missing from column (1). Parallel to Table 3.3, Table 3.4 lists colleges that meet our definitions of nontransferable voc-ed specialization and emphasis. Column (1) of Table 3.4 reports that just 10 colleges specialize in nontransferable voc-ed, meaning that they emphasize nontransferable voc-ed credits while de-emphasizing transferable credits. As in Table 3.3, _____________ 3In comparison, Santa Barbara City College and Los Angeles Trade-Tech are midsized with 14,913 and 15,630 students, respectively. Several colleges have more than 30,000 students, with Riverside Community College the largest at 34,042 students. 32 Table 3.3 Colleges Emphasizing and Specializing in a Transferable Curriculum, Alternative Definitions of Specialization Transfer Specialization Emphasis Based on Definition 1 (1) Chabot Coastline Cuyamaca El Camino Grossmont Hartnell Las Positas Los Angeles Harbor Los Angeles Pierce Marin Orange Coast Palomar Reedley San Diego Mesa San Francisco City San Joaquin Delta Sierra Skyline Southwest Definition 2 (2) Chabot Coastline Consumnes River Cuyamaca Cypress Diablo Valley El Camino Grossmont Hartnell Las Positas Los Angeles Harbor Los Angeles Pierce Marin Napa Valley Orange Coast Palomar Reedley San Diego Mesa San Francisco City San Joaquin Delta San Mateo Santa Barbara Santa Monica Sierra Skyline Southwestern CPEC Descriptions (3) Chabot Diablo Valley Fullerton Las Positas Los Angeles Harbor Los Angeles Pierce Marin Oxnard Santa Barbara Santa Monica Ventura NOTES: Definition 1 transfer specialization is an emphasis on transfer credits coupled with a de-emphasis on nontransferable voc-ed credits measured relative to all credits. Definition 2 transfer specialization is an emphasis on transfer credits coupled with a de-emphasis on nontransferable voc-ed credits measured relative to all voc-ed credits. 33 Table 3.4 Colleges Emphasizing and Specializing in Nontransferable Voc-Ed, Alternative Definitions of Specialization Nontransferable Voc-Ed Specialization Emphasis Based on Definition 1 (1) Allan Hancock Los Angeles Trade-Tech Los Medanos Monterey Peninsula Palo Verde Rio Hondo Santa Ana Santiago Canyon Taft Victor Valley Definition 2 (2) Allan Hancock Contra Costa Evergreen Imperial Valley Laney Los Angeles City Los Angeles Trade-Tech Los Medanos Monterey Peninsula Moorpark Oxnard Palo Verde Porterville Rio Hondo Santa Ana Santiago Canyon Taft Ventura Victor Valley CPEC Descriptions (3) Coastline Contra Costa Copper Mountain Consumnes River Desert East Los Angeles Foothill Fresno City Lassen Long Beach City Los Angeles Trade-Tech Moorpark San Diego Miramar San Joaquin Delta Shasta Siskiyous Taft NOTES: Definition 1 nontransferable voc-ed specialization is an emphasis on nontransferable voc-ed credits measured relative to all credits coupled with a de-emphasis on transfer credits. Definition 2 nontransferable voc-ed specialization is an emphasis on nontransferable voc-ed credits measured relative to all voc-ed credits coupled with a deemphasis on transfer credits. there is limited overlap between these 10 colleges and the 17 colleges listed in column (3) as having a voc-ed emphasis based on CPEC descriptions. Los Angeles Trade-Tech and Taft College appear in both columns, but Palo Verde College is listed in column (1) but not in column (3). 34 Correlation Result 2: Transferable Curriculum vs. Proportion of Voc-Ed That Is Nontransferable Making use of the detail available on the mix of voc-ed credits, we obtain a negative correlation coefficient (–0.766) between transferable credits to all credits and narrow voc-ed credits to all voc-ed credits.4 This negative relationship is reported in row 4 and column (1) of Table 3.2 and shown graphically in Figure 3.4. As in Figure 3.3, note in Figure 3.4 the concentration of data points in the upper left and lower right quadrants. A difference between the two figures is that the more narrowly defined denominator in the ratio of nontransferable voc-ed credits to all voc-ed credits has the effect of “spreading” horizontally the scatter diagram in Figure 3.4. Employing the imaginary box applied in Figure 3.3 with dimensions of 5 percentage points in each direction from the origin, we apply the same definitions of what it means to specialize in, say, a transferable curriculum by being above the mean of the transferable credits ratio by at least 5 percentage points while being below the nontransferable credits ratio by at least 5 percentage points. The greater horizontal spread of Figure 3.4 results in a larger number of colleges that meet this definition of specializing (termed “Definition 2”). Names of the 26 colleges that specialize in a transferable curriculum according to Definition 2 appear in column (2) of Table 3.3. Note that both SBCC and Santa Monica College are listed. Column (2) of Table 3.4 lists the 19 colleges that specialize in a nontransferable voc-ed curriculum according to the information included in Figure 3.4. Our preferred definition of specialization is the more inclusive definition implemented in column (2) of both tables. The reason is that the column (2) definition incorporates the breadth of Goal 3 PFE data covering transferable, voc-ed, and basic skills curriculums with the detail on voc-ed offerings available in Goal 4 PFE data. _____________ 4As described in Chapter 2, PFE Goal 4 data on the mix of voc-ed credits allow us to measure nontransferable voc-ed credits as a proportion of all voc-ed credits. 35 Deviations of transferable credits to all credits .4 .2 0 –.2 –.4 –.4 –.2 0 .2 .4 Deviations of nontransferable voc-ed credits to all credits .6 Figure 3.4—Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits Correlation Result 3: Positive Relationships Between Measures of Advanced Voc-Ed Curriculum Adding VTEA data for courses and programs to our PFE measures of credits, we obtain a cluster of three positive correlation coefficients relating variables measuring mix of advanced voc-ed credits, mix of advanced voc-ed courses, and advanced voc-ed courses to programs. The three positive correlation coefficients, which appear in rows 6 and 7 and columns (5) and (6) of Table 3.2, range in value from 0.598 to 0.802. Correlation Result 4: Tradeoff Between Student Interests in Voc-Ed and Transferring Our final correlation result introduces FTF measures of freshmen student academic interests. We find a sizable negative relationship between student interest in voc-ed and student interest in transferring. As seen in row 9 and column (8) of Table 3.2, the correlation coefficient between these two variables is –0.494. It is worth noting from row 9 that student interest in voc-ed is also negatively correlated with measures 36 of academic transfer credits (-0.202), advanced voc-ed courses (-0.287), and advanced courses to advanced programs (-0.236). Bringing It Together Using Factor Analysis The findings in the previous section are based on an analysis of measured relationships between nine of our curriculum variables. In this section, we buttress these findings with results obtained using a more rigorous, formal approach called factor analysis, which we apply to all 21 of our curriculum variables.5 Briefly, the objective of factor analysis is to help researchers categorize their data. More specifically, factor analysis identifies a small number of underlying unobserved variables, called factors, that account for measured relationships between a larger number of observed variables. Our results indicate that the relationships between our curriculum mix variables can be captured by just four unobserved factors. Ranked in descending order by their ability to distinguish colleges in terms of observed curriculum measures, the four factors and the labels we assign to each are as follows: Factor 1: transferable curriculum/nontransferable voc-ed tradeoff; Factor 2: advanced voc-ed curriculum; Factor 3: demand for nontransferable curriculum; and Factor 4: apprenticeship training. Allowing us to identify these four factors is an underlying matrix of factor loadings that we estimate to measure the importance of each observed variable in explaining a particular factor. (A detailed description of factor analysis and our key results is presented in Appendix B.) Beginning with Factor 1, the message we take from our results is a distinction between colleges that emphasize a transferable curriculum _____________ 5We actually used 18 of our 21 curriculum variables in the factor analysis. The three omitted variables are residual credits to all credits, first-time freshmen undecided about their academic objectives to all freshmen, and broadly defined voc-ed credits to all credits. These variables are omitted because they are linear combinations of the remaining variables. 37 and colleges that emphasize nontransferable voc-ed and appeal to students interested in basic skills. Large positive factor loads are obtained for measures of transferable credits to all credits and academic transfer credits to all credits. Negative factor loads appear for nontransferable voc-ed credits to all credits, nontransferable voc-ed credits to all voc-ed credits, and student interest in basic skills. In other words, Factor 1 appears to capture the two curriculum specializations we described in the previous section based on our Correlation Results 1 and 2. The first of these is a transfer curriculum specialization, and the second is specialization in nontransferable voc-ed. We label Factor 2 advanced voc-ed curriculum, as our data seem to be contrasting colleges that emphasize an advanced vocational curriculum with all other colleges. Large and positive factor loadings are observed for three variables: advanced voc-ed credits to all voc-ed credits, advanced occupational courses to all voc-ed courses, and advanced occupational courses to advanced occupational programs. Recall that positive correlation coefficients between these three variables were noted above as Correlation Result 3. Positive factor loads for Factor 3 are obtained for FTF measures of student interest in voc-ed, student interest in more broadly defined nontransferable programs (i.e., interest in voc-ed plus interest in an A.A. but not in transferring), and proportion of credits in basic skills. At the same time, a negative factor load appears for student interest in transfer programs. Since the variables with large factor loads are primarily calculated from FTF measures of student objectives, we label Factor 3 demand for nontransferable curriculum. Consistent with Correlation Result 4, these results suggest a distinction between colleges whose students express an interest in nontransferable programs and colleges whose students are interested in other curriculums, both voc-ed and traditionally academic, that are transferable to a four-year college. Factor 4 is clearly apprenticeship training. Large and positive factor loadings are observed for our measures of apprenticeship courses to all voc-ed courses, apprenticeship credits to all voc-ed credits, and apprenticeship programs per 100 students. Estimated factor loadings can be used to calculate predicted values of the common factors, called factor scores, for each community college in 38 our dataset. Thus, for each factor, factor scores allow us to measure the magnitude of differences between individual colleges. Table 3.5 lists the top five and bottom five colleges ordered by factor scores for each of the four factors. Beginning with Factor 1, the transferable curriculum/ nontransferable voc-ed tradeoff, both top-ranked Orange Coast College and bottom-ranked Taft College were mentioned in Chapter 2 as being at the extremes of the distribution for the PFE measure of transferable credits. All five top-ranked colleges appear in the Table 3.3 list of colleges that specialize in a transfer curriculum, using either the column (1) or column (2) definitions. Only Lassen College among the five bottom-ranked colleges does not appear in the Table 3.4 list of colleges that specialize in nontransferable voc-ed. With respect to the remaining factors, it is interesting to note for Factor 2 that Santa Barbara City College and Santa Monica College rank in the top five in terms of offering advanced voc-ed, although they do not rank in the top five for specializing in a transferable curriculum. The Table 3.5 Top Five and Bottom Five Colleges Ordered by Factor Scores for Factors 1–4 Factor 1: Transferable Curriculum/ Nontransferable Voc-Ed Tradeoff Orange Coast Palomar San Diego Mesa Sierra Los Angeles Pierce Taft Palo Verde Lassen Los Medanos Rio Hondo Factor 3: Factor 2: Demand for Factor 4: Advanced Voc-Ed Nontransferable Apprenticeship Curriculum Curriculum Training Top Five Colleges Cuesta Los Angeles Santiago Canyon Santa Barbara Southwest San Mateo Galivan Los Angeles Trade- Los Angeles Trade- Santa Monica Tech Tech Rio Hondo Los Angeles Mission Rio Hondo San Jose City Foothill Compton Bottom Five Colleges Santiago Canyon Taft Taft Marin Diablo Valley Siskiyous Merced Orange Coast Lassen West Hills DeAnza Canyon Palo Verde West Valley Merritt 39 top-ranked college for Factor 2, Cuesta College, was noted in Chapter 2 as being at the extreme right-hand tail of the distribution of advanced courses per advanced program. Summary Do California community colleges specialize in the academic programs they offer? Our answer, using three different empirical approaches, is that many do. We first examined interquartile ranges for our 21 curriculum mix variables to check whether large differences identified at the extremes of the distributions hold up more broadly across colleges. Our results indicated sizable differences between the top 25 percent and the bottom 25 percent of colleges for eight variables that are primarily measures of advanced level or transferability of voc-ed curriculums. Next we examined relationships between pairs of curriculum variables to determine whether intercollege differences can be interpreted as specialization. Using scatter diagrams to visualize pair-wise correlation coefficients between variables, we implemented a common-sense definition of specialization linking emphasis on one curriculum measure with a de-emphasis on another. Our preferred specification indicates that 26 colleges offer a transfer specialization, and 19 colleges specialize in nontransferable voc-ed. The correlation coefficients also suggested (1) a direct relationship between various measures of advanced voc-ed, and (2) a negative relationship between student interest in voc-ed and in other types of programs. Our third empirical approach used factor analysis to identify unobserved common factors that underlie observed relationships among all of our curriculum measures. We found that our data can be summarized by four common factors. Of the four factors, Factors 1-3 identify differences between colleges that reinforce three important results coming out of our pair-wise correlations. These are that (1) some colleges specialize in offering a transferable curriculum and others specialize in nontransferable voc-ed, (2) colleges differ in their emphasis on advanced voc-ed, and (3) colleges differ in the interest of their students in nontransferable voc-ed programs. 40 4. Explaining Differences in Curriculum Mix Chapter 3 presented evidence suggesting that California community colleges differ in terms of their curriculum emphases and specializations. In this chapter, we ask whether these differences can be explained. To answer this question, we append to our dataset measuring college curriculum variables additional data measuring college-specific characteristics, local job opportunities, and community demographics. The first section provides a brief overview of our measures of collegespecific characteristics, local job opportunities, and demographic characteristics. Appendix C includes a description of the online data sources and technical details involved in constructing these variables. A complete listing of all variables included in our dataset appears in Appendix D. Our main substantive results are reported in the second section, where we present cross-tabulation evidence indicating that differences between colleges in curriculum emphasis and specialization can be reasonably explained by our measures of college-specific characteristics and community needs. Description of Explanatory Variables College-Specific Characteristics Measures of campus-specific characteristics are broken up into two categories that we term institutional variables and student demographics. For the first category, Table 4.1 presents simple descriptive statistics for measures of membership in a multicampus district, proximity to a state four-year college, and campus age. Of the 108 CCCS campuses, 56 campuses (51.9 percent) are in 21 multicampus districts. The number of campuses included in each of these multicampus districts is not reported 41 Table 4.1 Descriptive Statistics for Explanatory Variables Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Age of campus (years) Student demographics Asian Black Filipino Hispanic Native American White Nonresident alien Other Nonresponse Female Male Gender unknown Employer characteristics Percentage of employees in Manufacturing Wholesale trade Retail trade Real estate Professional services Administration and support services Education services Health care Arts and entertainment Accommodations and food services Other services (except public administration) Total employees (thousands) Community demographics Bachelor’s degree Foreign born Standard Mean Deviation Minimum Maximum .519 — .000 1.000 47.5 51.4 26.9 30.9 53.1 22.6 1.7 282.5 1.5 178.2 1 119 .098 .090 .077 .097 .031 .028 .253 .156 .011 .009 .438 .199 .013 .014 .017 .015 .062 .047 .558 .062 .435 .064 .006 .010 .010 .003 .002 .039 .002 .021 .000 .000 .000 .200 .318 .000 .424 .694 .159 .856 .064 .869 .091 .178 .237 .677 .798 .054 .165 .123 .083 .057 .216 .087 .030 .020 .074 .048 .108 .063 .005 .004 .106 .049 .023 .029 .151 .064 .000 .002 .030 .008 .011 .013 .000 .009 .000 .030 .533 .400 .436 .179 .218 .322 .030 .256 .179 .398 .039 .014 76.3 178.4 .265 .146 .233 .116 .000 .085 .3 898.9 .054 .781 .013 .544 42 Variable Median household income ($ thousands) Race White Black Native American Asian Other race Two or more races Hispanic ethnicity Table 4.1 (continued) Standard Mean Deviation Minimum Maximum 49.3 21.2 21.9 173.6 .621 .167 .063 .075 .014 .008 .112 .110 .142 .104 .048 .013 .281 .174 .167 .002 .002 .003 .005 .019 .022 .937 .403 .045 .618 .494 .101 .893 in the table. The Los Angeles district is by far the largest district with nine campuses. The Peralta district serving the Oakland metropolitan area is second largest with four campuses. The next two rows of Table 4.1 measure proximity to the nearest UC and CSU campuses. The table indicates large variation about the means of 47.5 miles to the nearest UC campus and 26.9 miles to the nearest CSU campus. At the extreme of easy access, it is only 1.7 miles from Irvine Valley College to the nearest UC campus, and only 1.5 miles from San Francisco City College to the nearest CSU campus. At the other extreme, College of the Redwoods on the Northern California coast in Eureka is nearly 283 miles from the nearest UC campus, and Palo Verde College in the southeastern California desert, is 178 miles from the nearest CSU campus. Recall that Palo Verde College was described in Chapter 3 as a very small college that places heavy emphasis on nontransferable voc-ed. Campus age measured from 2002 is widely dispersed, but there is some concentration in the data at founding dates in the early to mid1950s. Chaffey College, founded in 1883, is the oldest CCCS institution; Copper Mountain College, founded in 2001, is the youngest. The student demographic variables in Table 4.1 include race and ethnicity as well as gender. Kane and Rouse (1999) report that at the national level, the combined student body of community colleges is 70 percent white, 11 percent black, and 11 percent Hispanic. (Kane and 43 Rouse do not break out the percentage of Asian students.) It is apparent that California community colleges enroll on average a much higher percentage of Hispanic students (25.3 percent) and a much lower percentage of white students (43.8 percent) than do community colleges nationally. Indeed, nine campuses mostly in the Los Angeles Basin and south to San Diego fall into the category of “predominantly Hispanicserving institutions” (PHIs), defined as those whose student bodies are over 50 percent Hispanic. A total of 47 community colleges, including the PHIs, fall into the more numerous “Hispanic-serving institutions” (HSIs) category, defined as those whose student bodies are 25 percent or more Hispanic. Imperial Valley College near the Mexican border reports the maximum proportion of Hispanic students (85.6 percent). Regarding gender, the unexpectedly low minimum ratio of female students (just 20 percent) is observed for Taft College. Local Job Opportunities Shown next in Table 4.1 are descriptive statistics for our two measures of local job opportunities—industry mix and total employment. Grubb (1996) emphasizes that the labor market for community college students (what he calls the “sub-baccalaureate labor market”) is almost entirely local. To concentrate on local labor markets, we make use of 1997 Economic Census information for employers in the immediate proximity of the community college campus using a threestep protocol described in Appendix C. With this protocol, we obtained unique college-specific employment information for 94 colleges. The remaining 14 colleges are all in large metropolitan areas, including Los Angeles, Oakland, Sacramento, San Diego, and San Jose. We measure employer characteristics using data for the entire city for at least two colleges in each of these metropolitan areas. In terms of industry mix, Table 4.1 shows that the largest spread between maximum and minimum values appears for manufacturing, followed by retail trade, wholesale trade, and accommodations and food services. Also observed is a huge variation in the size of the local labor market as measured by total employment. Total employment ranges between 335 employees for Foothill College in Los Altos Hills and nearly 44 900,000 employees for the four Los Angeles area colleges for which we use data for the Los Angeles metropolitan area. Community Demographics The community demographic variables appearing at the bottom of Table 4.1 are based on 2000 Census data for the city in which the college is located using the protocol developed for measuring local employer characteristics. The variation shown in proportions of bachelor’s degree holders and foreign-born residents and in median household income illustrates the diversity of California’s population. Maximum and minimum values for bachelor’s degree are obtained for Foothill College (78.1%) and Los Angeles Mission College (5.4%), respectively. The maximum value of median household income is also reported for Foothill College (over $173,000). The proportion of foreign-born residents ranges between the minimum reported for Lassen College (1.3%) and the maximum observed for Glendale Community College (54.4%). Lassen College is in the Sierra Nevada region of northeastern California; Glendale Community College is in the greater Los Angeles metropolitan area. Our remaining community demographic variables include race and ethnicity, each measured separately. The race/ethnicity mix of students may differ somewhat from the race and ethnicity of the community because colleges draw students from outside the geographic boundaries of their communities. Nevertheless, there appears to be a direct correspondence between the two. As one example, Feather River College reports the maximum proportion of white residents in its local community (93.7%) and the maximum proportion of white students (86.9%). Maximum values for other major race and ethnicity variables measured at the community level are 40.3 percent for blacks at Compton College, 61.8 percent for Asians at East Los Angeles College, and 89.3 percent for Hispanics at Los Angeles Mission College. The correlation between the racial and ethnic composition of students and community race and ethnicity is strong: We estimate correlation coefficients between corresponding student and community variables of 0.76 for blacks, 0.69 for Asians, 0.76 for Hispanics, and 0.76 for whites. 45 Explaining Differences in Specializations and Emphases We turn now to the question of the extent to which these explanatory variables may be helpful in understanding intercollege differences in curriculum specializations and emphases. Specializations in Transferable Curriculums and Nontransferable Voc-Ed Our two curriculum specialization measures are developed from a scatter diagram relating the ratio of transferable credits to all credits to the ratio of voc-ed credits that are nontransferable (see Figure 3.4). Colleges that specialize in a transferable curriculum are those above the mean of the transferable credits ratio by at least 5 percentage points but below the mean of the nontransferable voc-ed credits ratio by at least 5 percentage points. Similarly, colleges that specialize in nontransferable voc-ed are those above the mean of the nontransferable voc-ed credits ratio by at least 5 percentage points but below the mean of the transferable credits ratio by at least 5 percentage points. Column (1) of Table 4.2 displays means of selected explanatory variables for the 26 colleges listed in Table 3.3 as having a transfer curriculum specialty using Definition 2. Column (2) provides the same information for the 19 colleges listed in Table 3.4 that specialize in nontransferable voc-ed. Means of explanatory variables for the remaining 63 colleges appear in column (3). The first group of explanatory variables shown in Table 4.2 consists of campus-specific institutional variables. We expect to see colleges in multicampus districts offering a more specialized curriculum than colleges in single-campus districts. A college in a multicollege district must compete with other colleges to enroll local students, and one way to compete is for the college to differentiate itself in terms of its mix of academic programs. On the other hand, colleges that are the sole provider within reasonable commuting distance do not face this kind of competition and may be expected to offer a broad range of educational services to satisfy the diverse needs of local residents and employers. 46 Table 4.2 Cross-Tabulations Between Specializations in Transferable Curriculums and Nontransferable Voc-Ed and Selected Explanatory Variables Explanatory Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Student demographics Asian Black Hispanic White Employer characteristics Manufacturing employment Total employees (thousands) Community demographics Bachelor’s degree Foreign born Median household income ($ thousands) Race White Black Asian Hispanic ethnicity Transfer Specialization (1) Nontransferable Voc-Ed Specialization (2) Other Colleges (3) .577 .632 .460 25.0 54.7 54.6 15.6 33.9 29.4 .120 .082 .094 .058 .081 .084 .219 .343 .239 .451 .359 .457 .159 86.1 .185 128.9 .152 56.5 .299 .204 .269 .250 .264 .217 51.6 46.1 49.3 .649 .564 .627 .043 .074 .068 .116 .081 .120 .265 .411 .248 Number of colleges 26 19 63 Table 4.2 demonstrates that, consistent with these expectations, colleges that specialize are more likely than other colleges to be part of multicampus districts. In particular, columns (1) and (3) indicate that colleges with a transfer specialization are nearly 12 percentage points more likely to belong to a multi-campus district than are “other” colleges. Similarly, columns (2) and (3) show that colleges with a nontransferable voc-ed specialization are 17 percentage points more likely to belong to a multicampus district than are other colleges. 47 Consider next proximity to a four-year college. For at least two reasons, colleges close to a four-year college are more likely than other colleges to specialize in a transferable curriculum. First, proximity makes transferring cheaper for students because they may be able to continue to live with their parents or in their own homes. Second, proximity is likely to lead to closer relationships between faculties of the community college and the four-year college, which, in turn, should result in a greater number of articulation agreements and a more seamless transfer process. Again, results shown in Table 4.2 are consistent with expectations. A comparison of columns (1) and (2) shows that colleges that specialize in a transferable curriculum are on average nearly 30 miles closer to a UC campus and about 18 miles closer to a CSU campus than are colleges with a nontransferable voc-ed specialization. Similarly, columns (1) and (3) show that colleges with a transferable curriculum specialization are over 29 miles closer to a UC and about 14 miles closer to a CSU than are other colleges. In other words, colleges offering a transferable curriculum are on average about half the distance to the nearest UC or CSU campus compared to colleges with a nontransferable voc-ed specialization and all other colleges. Within the other college category, we singled out for further analysis 17 colleges offering what might be termed a “comprehensive” curriculum, by which we mean a mix of transferable credits and nontransferable voc-ed credits that is within 5 percentage points of the mean of each variable. Probability of membership in a multicampus district is 41 percent for these colleges compared to the 46 percent shown in Table 4.2 for all 63 other colleges. Consequently, gaps in multicampus membership between colleges with a curriculum specialization and comprehensive colleges are even larger than the gaps appearing in the table. For example, colleges with a transfer specialization are nearly 17 percentage points more likely than comprehensive colleges to belong to a multicampus district, compared to the 12 percentage point gap in Table 4.2 between colleges with a transfer specialization and other colleges. Similarly, comprehensive colleges are even farther away from the nearest UC and CSU campuses than are the 63 other colleges—about 65 miles to the nearest UC and 39 miles to the nearest CSU. 48 Regarding race and ethnicity of students, Table 4.2 suggests that Asian students are slightly more likely to attend colleges with a transfer specialization than colleges specializing in nontransfer voc-ed. On average, Asian students represent 12 percent of students attending colleges with a transfer specialization, whereas 8 percent of students attending colleges with a nontransfer voc-ed specialization are Asian. Among the other race and ethnicity categories, columns (1) and (2) indicate a larger difference in the opposite direction for Hispanic students. Hispanics are over 12 percentage points more likely to attend a college specializing in nontransferable voc-ed than to attend a college with a transfer specialty. Turning to employer characteristics, Table 4.2 indicates that colleges specializing in nontransferable voc-ed tend to be in larger labor markets than are colleges with a transfer specialty. In addition, both categories of colleges that specialize are in larger labor markets than are the remaining 63 colleges. Differences in mix of employment, in contrast, tend to be quite small. The table shows that the percentage of local employment in manufacturing is higher by less than 3 percentage points for colleges that have a nontransferable voc-ed specialty than it is for colleges with a transfer specialty. Finally, the community demographic variables shown in Table 4.2 suggest that colleges with a transfer specialization are somewhat more likely to be in more highly educated, higher-income communities than are colleges with a nontransferable voc-ed specialization. For example, columns (1) and (2) indicate a nearly 10 percentage point difference in the percentage of residents with a bachelor’s degree. Echoing an earlier result for Hispanic students is the finding that communities with colleges that specialize in nontransferable voc-ed have a 15 percentage point greater share of Hispanics than do communities with colleges with a transfer specialization. Emphasis on Advanced Voc-Ed In Chapter 3, we noted as Correlation Result 3 the existence of large positive relationships between three measures of advanced voc-ed: (1) advanced voc-ed credits to all voc-ed credits, (2) advanced voc-ed courses to all voc-ed courses, and (3) ratio of advanced voc-ed courses to 49 advanced voc-ed programs. Any one of these variables would serve as a reasonable candidate for measuring an advanced voc-ed emphasis. The particular advanced voc-ed variable we examine in Table 4.3 is advanced voc-ed courses to all voc-ed courses. The first two columns of the table contrast colleges in the top and bottom quartiles of the ratio of advanced voc-ed courses to all voc-ed courses for the same selected explanatory variables appearing in Table 4.2. Table 4.3 Cross-Tabulations Between Emphasis on Advanced Voc-Ed Courses and Student Interest in Voc-Ed and Selected Explanatory Variables Explanatory Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Student demographics Asian Black Hispanic White Employer characteristics Manufacturing employment Total employees (thousands) Community demographics Bachelor’s degree Foreign born Median household income ($ thousands) Race White Black Asian Hispanic ethnicity Number of colleges Emphasis on Advanced Voc-Ed Top Quartile (1) Bottom Quartile (2) Student Interest in Voc-Ed Top Quartile (3) Bottom Quartile (4) .462 .333 .630 .667 31.9 73.5 35.7 39.5 17.6 41.6 27.4 19.6 .092 .070 .094 .123 .052 .078 .120 .054 .273 .227 .310 .200 .446 .513 .333 .482 .183 61.4 .124 93.3 .089 167.3 .092 50.8 .270 .210 .251 .333 .229 .199 .282 .232 49.9 40.3 52.1 54.0 .641 .630 .533 .680 .045 .080 .082 .037 .103 .077 .132 .126 .307 .276 .348 .211 26 27 27 27 50 Comparing columns (1) and (2) of Table 4.3, the primary differences between colleges in the top and bottom quartiles are for the campusspecific institutional variables. In particular, colleges emphasizing advanced voc-ed are 13 percentage points more likely than colleges that do not emphasize advanced voc-ed to be part of a multicampus community college district. In addition, top-quartile colleges for this variable are nearly 42 miles closer to a UC campus and 24 miles closer to a CSU campus than are bottom-quartile campuses. If we pull together comparable estimates in Tables 4.2 and 4.3, the picture that emerges is that colleges with a transfer specialization or an emphasis on advanced voc-ed are more likely to be part of a multicampus district and to be within reasonable commuting distance to a UC or CSU campus. Differences between colleges in emphasis on advanced voc-ed tend to be small for race and ethnicity, whether measured for students or for the community. However, the limited evidence in Table 4.3 suggests that emphasis on advanced voc-ed is somewhat more common for colleges in more highly educated, higher-income communities. For example, the incidence of bachelor’s degrees is 6 percentage points higher for top quartile than for bottom quartile colleges. Top quartile colleges also tend to be in somewhat smaller labor markets and in labor markets with a higher proportion of manufacturing employment. Student Interest in Voc-Ed The third difference between colleges singled out for further attention in Chapter 3 is in freshmen student interest in voc-ed programs. We also mentioned in Chapter 2 that student interest in voced is likely to be primarily capturing demand for nontransferable voc-ed. Columns (3) and (4) in Table 4.3 compare top and bottom quartiles of colleges measured in terms of student interest in voc-ed. It should be noted that colleges appearing in the bottom quartile, indicating a low level of student interest in voc-ed, are likely to be colleges whose students exhibit a high level of interest in transfer programs. A comparison of columns (3) and (4) in Table 4.3 indicates small differences between top and bottom quartile colleges for membership in a multicampus district and proximity to the nearest four-year college. 51 Larger differences between quartiles appear for student demographic variables. In particular, proportions of black and Hispanic students are 7 percentage points and 11 percentage points higher, respectively, for top quartile than for bottom quartile colleges. On the other hand, colleges with large proportions of Asian and especially white students are more likely to enroll students with relatively little interest in voc-ed. Race and ethnicity results at the community level echo these findings. For instance, the proportion of Hispanic residents is nearly 14 percentage points higher for top quartile than for bottom quartile colleges. Student interest in voc-ed is also positively related to proportion of foreign-born residents in the community and to the size of the labor market and negatively related to the proportion of bachelor’s degree holders (Table 4.3). Summary Colleges that specialize in either a transferable curriculum or in nontransferable voc-ed are more likely than other colleges to be part of a multicampus district and to be in large labor markets. Between these two categories of specialization, colleges that have a transfer specialization are in closer proximity to a UC or CSU campus, whereas colleges that specialize in nontransferable voc-ed tend to have a larger proportion of Hispanic students and to be in larger communities with greater minority populations. We also compared colleges that differ in their emphasis on advanced voc-ed curriculums and in the interest of their freshmen students in voced programs. Emphasis on advanced voc-ed is directly related to membership in a multicampus district, proximity to a UC or CSU campus, and proportion of manufacturing employment. The racial and ethnic mixes of the student body and the community have a stronger effect on student interest in voc-ed programs, with interest in voc-ed increasing with the proportions of Hispanics attending the college or residing in the community. Demand for voc-ed programs also increases with size of the local labor market. 52 5. Policy Discussion The purpose of this concluding chapter is to provide guidance to policymakers for evaluating the performance of community colleges. The chapter is divided into three sections. The first reviews key provisions of the California Master Plan for Higher Education. The second provides some detail on the multiple performance indicators faced by community colleges. The third discusses several observations drawn from our empirical findings that we suggest should be taken into account when evaluating California community colleges. Key Provisions of the California Master Plan The 1960 California Master Plan laid the foundation for the subsequent development of the California Community College System on two guiding principles. The first is a policy of open admission to state high school graduates at no tuition (later, that principle was modified to very low tuition). California still leads the nation in accessibility of community college educational services and low tuition (Romano, 2003). The second guiding principle is the flexibility granted to community college districts and individual colleges to define their own missions. As stated in the Master Plan, community colleges are to offer instruction in (1) standard collegiate courses for transfer to fouryear institutions, (2) vocational-technical fields leading to employment, and (3) general or liberal arts courses (California State Department of Education, 1960, p. 36). Hence, community college districts and individual colleges have for decades enjoyed considerable latitude in setting their missions in line with perceived local business and community needs and with the vision of college leaders. Over time, the Master Plan has frequently been amended by the California state legislature. An important amendment was passed in 1996, reiterating that the primary missions of California community colleges are to offer traditional academic programs and vocational 53 instruction at the lower-division level. Academic and voc-ed programs are to be accessible to both traditional college-age and older students, including adults returning to school. As part of these primary missions, the amendment specifies that colleges are to offer remedial instruction and adult noncredit education. The 1996 amendment states that an additional primary mission of community colleges is “to advance California’s economic growth and global competitiveness through education, training, and services that contribute to continuous workforce improvement” (State of California, 1996, section 2). To help understand the legislature’s intent, other provisions of the amendment add that in fulfilling this new mission, community colleges are to play a more active and even entrepreneurial role in local and regional economic development. Along with this emphasis on workforce development, the legislature also established in 1996 the California Performance Based Accountability (PBA) system to monitor the performance of community colleges in meeting workforce development goals. Performance Evaluation Requirements As their missions have expanded, California community colleges have also been faced with increasing demands for accountability in their use of taxpayers’ money. Chapter 1 described the federal student-rightto-know regulation requiring that colleges collect and report student transfer and completion rates. As we noted above, data for First-Time Freshman student cohorts are used to calculate SRTK completion and transfer rates. More important are the reporting requirements established by two additional pieces of federal legislation, both of which were passed in 1998. The Workforce Investment Act specifies a total of 17 core indicators of performance and consumer satisfaction for state institutions, including community colleges. Core indicators of performance include such labor market outcomes as placement in unsubsidized employment, initial wages, and retention in employment and wages measured after six months on the job. These employmentrelated measures are to be obtained from Unemployment Insurance (UI) quarterly earnings histories. Levels of performance on core indicators 54 that community colleges are expected to meet or exceed are established in negotiations between federal and state officials. Specific to voc-ed programs, VTEA establishes a similar set of core indicators of performance and requires that states collect and report data with which performance can be assessed. Not covered under WIA is an additional VTEA goal requiring completion of voc-ed programs that lead to nontraditional training and employment. Performance goals required by WIA were established in the 1996 Partnership for Excellence agreement between the State of California and the CCCS. The five PFE goals include • A greater number of transfers to UC and CSU campuses; • An increased number of degrees and certificates awarded; • Higher rates of course completion for transfer, voc-ed, and basic skills courses; • Greater contribution to workforce development as measured by completion of apprenticeship, advanced voc-ed, and introductory voc-ed courses; and • Basic skills improvement as measured by number of students completing coursework at least one level above prior basic skills courses. The core indicators required by federal VTEA legislation are specified in the 1994 California Perkins State Plan. These core indicators include • Completion rates for courses in voc-ed programs; • Percentages of voc-ed students who transfer to a UC or CSU campus, earn a degree or certificate, or join the military; • Placement and retention of voc-ed students in UI-covered employment; and • Percentages of underrepresented gender students participating in voc-ed programs leading to nontraditional employment. It is also worth noting that the federal WIA and VTEA acts require that data be collected by program area. Hence, at the state level, VTEA and PFE data are available for California community colleges by program. This level of detail allows college administrators to conduct 55 internal program evaluation. Similarly, students have the information necessary to make an informed comparison of different colleges that offer the same programs and of different programs offered by the same college. Guidance to Policymakers for Evaluating Community Colleges Given the number and variety of performance standards they face, community colleges must shoulder a substantial burden connected with collecting and reporting required information. Added to this burden is the possibility of failing to meet one or more of the performance criteria and losing state and federal funding. As might be expected in these circumstances, community college administrators are vocal in their criticism of existing performance standards and not at all reticent in offering suggestions for how these standards might be revised. These criticisms include the following: • The costs of collecting required data are high. Especially burdensome are the requirements of following former students for up to six months and collecting and reporting data by program for each institution. • The standards are poorly designed to capture the range of successful student outcomes given the range of student backgrounds and interests. • Standards fail to reflect the multiple missions of community colleges. • Standards are likely to have unintended and perverse consequences. These include the dropping of worthwhile courses that do not figure into the reward system, failing to admit students who are less likely to graduate or to be easily placed in jobs or in four-year colleges, and grade inflation to boost completion and retention rates. If carried to an extreme, these criticisms suggest that each community college should be evaluated individually in terms of its performance in meeting the needs of the local community. A reasonable evaluation instrument for this purpose would be focus group interviews, 56 and we mentioned that a WIA performance indicator is consumer satisfaction with training services received. Nevertheless, the clear intent of legislators at both the federal and state levels is to mandate accountability standards requiring minimum levels of performance that are applied consistently across all colleges. It is a difficult task to establish performance standards that insure minimum levels of performance while recognizing that community colleges may have different missions. Our objective in this report is not to provide specific evaluation guidelines. Rather, our more modest objective is to develop evidence that informs policymakers, helping them to establish evaluation standards that contribute to improved performance without being viewed by community colleges as onerous or even counterproductive. What Does Our Evidence Show? We believe that the following observations drawn from our empirical results will be helpful to policymakers in formulating standards used in evaluating community colleges. 1. On average, California community colleges offer more credits in transferable programs than in other broad curriculum categories. In Chapter 2, we discussed PFE data on the distribution of credits offered, broken down by transferable, nontransferable voc-ed, adult basic skills, and other credits; and we noted that, on average across CCCS campuses, colleges offer 73 percent of total credits in transferable programs. Further analysis in Chapter 3 indicated that most colleges offer at least half of all credits in transferable programs. Moreover, the interquartile range for the proportion of transferable to all credits is found to be quite narrow, lying between 69 percent and 79 percent. With a few exceptions, our evidence suggests that California community colleges are heavily engaged in offering transferable curriculums. 2. Transferable curriculums are a mix of voc-ed courses and traditional academic courses. Simply comparing transfer rates, as in a recent study by Ehrenberg and Smith (2004), may be misleading. Transferable curriculums are typically thought to be composed of traditional academic courses, but as we pointed out in Chapter 2, they are in fact a mix of credits generated by transferable voc-ed courses as well as traditional 57 academic courses. Broadly defined voc-ed that includes both transferable and nontransferable voc-ed credits is, on average across CCCS campuses, about 23 percent of all credits. Of all voc-ed credits, in turn, about 64 percent are on average transferable credits. In Chapter 2, we focused on Santa Barbara City College as an example of a CCCS college that promotes itself as having a strong transfer orientation. It is interesting to note that at 77 percent, the ratio of transferable voc-ed to all voc-ed credits for SBCC is even higher than the all-college average (73%). Because skills acquired in transferable voc-ed programs are likely to be valued in local labor markets, a successful outcome for a college with a strong transferable voc-ed orientation might well be placement in a training-related job as opposed to a successful transfer to a four-year college. As indicated by Grubb (1996), probably the best way to measure the labor market effects of voc-ed curriculums is to use matched data that link student records to UI wage records. At the federal level, WIA and VTEA mandate the use of merged files containing student records and UI wage records to calculate employment, wage, and job retention outcomes. In their recent WIA implementation report, however, Macro, Almandsmith, and Hague (2003, chapter 8) indicate that few states have data currently available to evaluate the performance of training providers. Thus, performance standards specified in terms of labor market outcomes by program are likely to require the funding of additional institutional research by community colleges. Short of this financial commitment, the state might consider taking on centrally the management of large and complex administrative datasets and the production of the required data. 3. Although most credits offered by most colleges are transferable, there are important differences across CCCS campuses in curriculum “emphasis” and “specialization.” Information on differences in the missions of community colleges may be obtained by reading colleges’ promotional materials or by interviewing college officials in site visits. Our primary approach, in contrast, is to measure differences in the curriculum mixes offered to students. We argue that important differences in these mixes may be captured empirically by differences in curriculum “emphasis” and “specialization.” What we mean by curriculum emphasis is systematic 58 deviations from the mean of a particular curriculum measure. As reported in Chapter 3, the data suggest that colleges differ in the emphasis they place on voc-ed offerings, especially in the mix of voc-ed credits and courses that are transferable or taught at an advanced level. We also find that colleges differ in terms of the interest of their freshmen students in transferable versus voc-ed programs. Our definition of curriculum specialization goes beyond emphasis to link emphasis on one type of curriculum with a de-emphasis on another. Making use of this definition, we find that some colleges offer a transfer specialization including both traditional academic and voc-ed curriculums, whereas others specialize in nontransferable voc-ed. The number of colleges that specialize on these two dimensions depends on exactly how we implement our definition of specialization, and for this we consider two alternatives. Using our preferred alternative, we find that 26 colleges have a transfer curriculum specialty and another 19 colleges specialize in nontransferable voc-ed. Viewing these differences in curriculum emphases and specializations as indicating important differences in missions, we conclude that a “one-size-fits-all” evaluation strategy may not be appropriate. 4. Basic skills programs are a small proportion of most community colleges’ total offerings. Employers frequently cite lack of basic skills as a major obstacle to finding qualified employees, and adult basic skills curriculums are closely linked to community colleges’ workforce development mission. In Chapter 2, we reported that on average about 11 percent of full-time freshman students indicate that their primary academic objective is to acquire basic skills. Measured in terms of credits, about 7 percent on average of all credits are offered in basic skills courses. Although all colleges report freshman student interest in basic skills and offer basic skills courses for credit, we found in Chapter 3 that variation across colleges in our measures of basic skills is limited, especially in terms of credits offered. Our data thus indicate that the mission to provide basic skills courses is not one that receives great prominence, nor is it one on which community colleges attempt to differentiate themselves. Note that this conclusion rests on the assumption that the primary business of community colleges is to offer courses that supply credits toward a degree. 59 5. Colleges are responsive to differences in community characteristics in the curriculum mixes they offer. In Chapter 4, we examined the question of whether differences across colleges in curriculum emphases and specializations can be explained by indicators of college-specific characteristics and measures of community needs. College-specific characteristics include proximity to a four-year college and membership in a community college district, whereas community needs are measured by characteristics of the local labor market and community demographics. Our evidence indicates clearly that differences between colleges in curriculum mix are not randomly determined. In terms of collegespecific characteristics, we find that curriculum specializations and emphases have the expected relationships with membership in a multicampus district and proximity to a UC or CSU campus. Similarly, these measures of specializations and emphases appear to be related to the racial and ethnic composition of both students and the local community, in addition to other community demographic characteristics such as proportion of residents holding a bachelor’s degree. The conclusion we draw from this evidence is that California community colleges are responding in expected ways to differences in the interests and backgrounds of their students and to differences in the needs of their communities. In terms of accurately evaluating performance, these differences in missions should be taken into account. Allowing for Differences in Missions Precisely how to allow for differences in missions in measuring performance is a difficult issue that is currently receiving attention by policymakers in states across the nation. We conclude our report by commenting briefly on two possible approaches. The first was provided in the Job Training Partnership Act, which regulated federal employment and training activities for most of the 1980s and 1990s. The model-based JTPA procedure allowed performance standards to differ across colleges in response to quantitative differences in student characteristics and local economic conditions. Our evidence suggests that the model-based approach has merit. We find that differences in student characteristics and local economic conditions are closely related 60 to differences in colleges’ missions, as measured by differences in curriculum mix. Nevertheless, the model-based JTPA adjustment procedure was dropped from the Workforce Investment Act in 1998. In its place, WIA substituted a more ad hoc procedure that permits states either to waive performance standards or to negotiate adjustments in expected levels of performance. Such adjustments may be permitted if a college can make the argument that “unanticipated circumstances” involving a change in economic conditions or student characteristics make performance standards unreasonably burdensome. Analysts such as Barnow and Smith (2004) suggest that a fairer and less ad hoc procedure would be to return to the model-based system used under JTPA. A second approach involves up-front recognition by the state that community colleges differ in their missions. That is, rather than adjusting levels of performance expected for a given set of accountability standards, community colleges would be allowed to report data for those performance standards deemed to be consistent with their missions. There is some evidence that this approach may be receiving consideration at the federal level. First, the Employment and Training Administration (ETA) of the U.S. Department of Labor recently recommended reducing the 17 core WIA performance measures to a set of eight “common” measures, four of which would apply to adult programs and four to youth programs (Employment and Training Administration, 2003). Second, the U.S. House of Representatives, building on a Bush administration proposal, recently crafted and passed a WIA reauthorization bill (H.R. 1261). The reauthorization bill follows the ETA recommendation to reduce the number of performance criteria while giving governors authority to modify statewide standards in response to recommendations of local workforce investment boards and training service providers. Should California policymakers choose to take this approach to performance evaluation, we hope that the methodology developed in this report is helpful in measuring differences across community colleges in their missions. 61 Appendix A Technical Detail on Measures of Curriculum Mix PFE Data on Enrollment Partnership for Excellence data provide breakdowns of student credit hours attempted and completed by (1) type of course categorized by transferable, voc-ed, and basic skills, and (2) type of voc-ed program categorized as apprenticeship, advanced occupational, and introductory occupational. In the first or “Goal 3” breakdown, transferable courses are defined as courses that are transferable to UC or CSU campuses. Voc-ed courses, in contrast, are defined as nontransferable courses with Student Accountability Model (SAM) Priority Codes A, B, and C, classifying courses as “apprenticeship,” “advanced occupational,” and “clearly occupational,” respectively. The basic skills classification includes courses that supply precollege basic skills and basic skills unrelated to college entrance. In the PFE dataset, basic skills courses are assigned SAM Priority Codes D and E, indicating that the courses are “possibly occupational” and “nonvocational,” respectively. The Goal 3 breakdown of credits has the advantage of supplying a comprehensive overview of credits, but it provides a “narrow” specification of voc-ed. The second or “Goal 4” breakdown of voc-ed course enrollment retains the Goal 3 restriction limiting voc-ed courses to those with SAM Priority Codes A, B, and C. However, voc-ed courses are not restricted to nontransferable courses, thus allowing a “broad” specification of voced that includes transferable courses. The curriculum mix variables described in Table 2.2 include both narrow and broad measures of voc-ed. 63 VTEA Data Measuring Courses and Programs The federal VTEA and California state law require that each community college voc-ed program is to be evaluated using four core indicators of performance. For this purpose, the CCCS Chancellor’s Office maintains an electronic master Taxonomy of Programs (TOP) code file that lists over 300 individual programs and assigns to each a unique TOP code number. Vocational courses are defined as those having SAM Priority Codes A, B, C, or D and a vocational TOP code. Our data measuring voc-ed courses come from VTEA “Core Indicator Reports” using the “Course Voc Status/SAM Code” file. This file provides the number of voc-ed course offerings with SAM Codes A, B, C, and D for each community college in 1998–1999. In Table 2.3, the curriculum mix variables we report for voc-ed courses are constructed using data for SAM Codes A, B, and C. Total voc-ed courses are defined as the sum of SAM Codes A, B, and C courses. We use the same VTEA “Course Voc Status/SAM Code” file to measure the number of voc-ed programs offered grouped by SAM Priority Codes A, B, and C. Courses in each SAM Priority Code are listed by a TOP code. Our procedure for arriving at a number of programs in each priority code is to go down the list of courses offered and then to sum up the number of courses with different TOP codes. Two points should be noted regarding these data. First, since TOP codes are matched to courses, what we measure are program “areas” defined by content, which may not be the same as programs officially described in college catalogs. Second, our measures of SAM Codes A, B, and C programs are not mutually exclusive. The reason is that a program will often include courses categorized at more than one SAM Code level. For example, a program with a particular TOP code number might include advanced occupational (SAM Code B) courses as well as beginning occupational (SAM Code C) courses. Hence, this TOP-coded program would be counted in our dataset as both a SAM Code B program and a SAM Code C program. Chapter 2 notes that we do not have a reliable measure of total voc-ed programs with VTEA data. 64 Program Inventory Data In addition to VTEA data on voc-ed programs offered, the Chancellor’s Office maintains an “Inventory of Approved and Projected Programs.” The program inventory dataset includes all programs that are “approved” in the sense that a program must require 18 or more credits of course work. Hence, the number of programs included in this dataset is likely to understate the number of programs described in college catalogs, whereas, as noted, VTEA data report program areas measured by content rather than programs as they are officially titled. Each approved program is listed in the program inventory data by TOP code, and the Chancellor’s Office provides a list of programs matched to TOP codes that singles out vocational programs. Total approved voc-ed programs are also provided. First-Time Freshman Data As described in Chapter 2, FTF data are derived from a longitudinal study following the universe of first-time freshmen at all CCCS campuses between their initial enrollment in Fall 1997 and Spring 2000. Detailed information available by campus is presented in three categories: cohort, awards granted, and transfers. Cohort information includes gender, age, and race/ethnicity of students. In addition, FTF cohort data include a breakdown of freshman students’ academic objectives uninformed by counseling (“uninformed goals”) and informed by counseling (“informed goals”). Cohort data measuring students’ academic objectives uninformed by counseling are used in Chapter 2 to represent the demand of incoming students for alternative categories of educational services (voc-ed, transfer, basic skills, and other). FTF awards-granted data include total awards earned over the 1997–2000 period for the 1997 freshman cohort broken down by gender and race/ethnicity. Information is also available on type of award (A.A., A.S., and certificates) and title of program completed. Finally, FTF transfer data include total number of transfers broken down by whether the destination college is public or private, in California or out of state, the semester of transfer, and the name of the destination fouryear college or university. 65 Appendix B Technical Detail on Factor Analysis We implemented our factor analysis using as a guide the multistep strategy outlined by Johnson and Wichern (1988, p. 415). The software we used is the factor procedure in base SAS (SAS Institute, 1990). Johnson and Wichern recommend, as a first step, performing a principal component factor analysis with varimax (orthogonal) rotation. Our varimax rotation results yielded six common factors based on the criterion that the eigenvalues of the correlation matrix are greater than 1. Inspection of these results shows that the first four common factors extracted have a clear interpretation in terms of our curriculum mix variables. Moreover, each possesses the desirable feature of having at least three curriculum mix variables with large factor loadings. The remaining two common factors were both difficult to characterize and had fewer than three variables with substantial factor loadings. Since the last two common factors added little to our understanding, we restricted further estimations to four common factors. Factor loadings obtained using a varimax rotation for our 18 independent curriculum measures are reported in Table B.1. Continuing to follow the suggestions of Johnson and Wichern, we checked the sensitivity of our factor loadings to outliers, and we relaxed the orthogonality restriction by allowing for correlation among factors using the promax oblique (nonorthogonal) rotation. The main outlier we have identified is Santiago Canyon College with its heavy emphasis on apprenticeship training. Its omission caused only minor changes in the size of our factor loadings, so we retained Santiago Canyon in our analysis. Similarly, the promax rotation yielded results very similar to those obtained from the varimax rotation with little evidence of significant correlation between factors. 67 Table B.1 Factor Loadings from the Orthogonal Rotated Factor Patterns Curriculum Measure Transferable credits/ all credits Academic transfer credits/all credits Nontransferable voc-ed credits/all credits Nontransferable voc–ed credits/all voc-ed credits Interest in basic skills Advanced occupational credits/all voc-ed credits Advanced occupational courses/all voc-ed courses Advanced courses/ advanced programs Interest in voc-ed Interest in nontransfer programs Basic skills credits/ all credits Interest in transferring Apprenticeship courses/all voc-ed courses Apprenticeship credits/ all voc-ed credits Apprenticeship programs/100 students Voc-ed courses/100 students Advanced occupational programs/100 students Transferable Curriculum (1) .912 .801 –.906 –.711 –.677 .166 .289 –.019 –.053 .013 .041 .358 –.002 –.037 –.069 –.345 –.111 Advanced Voc-Ed Curriculum (2) .164 .050 –.105 –.176 –.044 .845 .841 .832 –.323 –.310 .069 .261 –.027 –.142 .168 .073 .195 Demand for Nontransferable Curriculum (3) –.060 –.230 –.149 –.128 .097 .138 –.083 –.108 .779 .777 .652 –.597 –.093 –.080 .114 .346 .413 Apprenticeship Training (4) –.096 .002 .149 .240 –.263 .022 –.031 .034 .230 .265 –.023 .051 .936 .862 .624 –.058 –.165 68 Table B.1 (continued) Curriculum Measure Approved voc-ed programs/total approved programs Variance explained (%) Transferable Curriculum (1) Advanced Voc-Ed Curriculum (2) –.018 24.5 –.029 13.2 Demand for Nontransferable Curriculum (3) –.073 12.7 Apprenticeship Training (4) .027 10.6 NOTES: Omitted curriculum variables are residual credits to all credits, undecided first-time freshmen to all freshmen, and broadly defined voc-ed credits to all credits; Copper Mountain College is omitted from the analysis. Factor loadings greater than 0.5 in absolute value are in bold. Finally, Johnson and Wichern suggest performing a maximum likelihood factor analysis including a varimax rotation. We were unable to successfully carry out the maximum likelihood analysis because the estimator gave communality estimates (squared multiple correlations of the common factors with the ith variable) that are greater than 1. This situation, described in the statistical literature as an “ultra-Heywood case,” is a particular problem when using the maximum likelihood method because variables with high communality are given high weights in subsequent iterations. The factor loadings shown in Table B.1 can be used to calculate predicted values of the common factors for each community college in our dataset. These predicted values are called factor scores. Descriptive statistics for the factor scores calculated for each of our four factors are shown in Table B.2. Note that factor scores are standardized to mean 0 and variance 1. Hence, the maximum/minimum and interquartile ranges shown in the table can be interpreted in terms of standard deviations. Colleges with the top five and bottom five factor scores for Factors 1–4 are listed in Table 3.5. 69 Table B.2 Descriptive Statistics for Factor Scores Obtained from Factor Analysis Common Factors Factor 1: transferable curriculum Factor 2: advanced vocational curriculum Factor 3: demand for nontransferable curriculum Factor 4: apprenticeship training Minimum Maximum –6.61 1.80 –1.36 4.50 –2.04 –1.37 3.04 7.77 Max/Min Range 8.46 IQ Range 0.89 5.86 1.23 5.07 1.23 9.14 0.57 NOTES: Factor scores are standardized to 0 mean and unit variance. Copper Mountain College is omitted from the analysis. 70 Appendix C Data Sources and Technical Details Used to Construct the Explanatory Variables College-Specific Characteristics Table 4.1 describes four college-specific variables. Beginning with multicollege district, PFE data on number of colleges in the district (Var9) are used to create a dummy variable indicating whether the college is part of a multicampus community college district. We used a three-step procedure to calculate proximity to the nearest UC or CSU campus. There are 10 campuses in the UC system and 23 campuses in the CSU system. Making use of a large map of California, we first identified at least two UC campuses and at least two CSU campuses as potentially nearest to each community college. For community colleges in large metropolitan areas such as Los Angeles and San Francisco, we picked up to seven UC campuses and seven CSU campuses for comparison. In the second step, we used Yahoo maps to calculate for each UC-CC pair and each CSU-CC pair (1) driving distance in miles and (2) travel time in minutes. The third step in the procedure involved choosing for each community college the nearest UC and the nearest CSU in driving distance and driving time. Proximity measured in miles is reported in Table 4.1. Campus age is obtained from the web site for each college. We strived to pin down the year the campus was founded, even if the college was originally a branch campus or subsequently changed its name. Student demographic data come directly from CPEC. 71 Local Job Opportunities Using online data from the 1997 Economic Census, we sought to obtain information about employers in the immediate proximity of the community college campus. Logically, this is best accomplished by selecting as the local labor market the city in which a community college is located. However, this characterization of the local labor market does not work for cities that are large enough to include multiple colleges. Nor does it work for communities that are too small to have the required Census data. In an attempt to deal with these problems, we used the following three-step protocol. 1. Check the city named in the college’s mailing address. If there is only one college in this city and Census data are available, use these data for this city. 2. If the city indicated in the mailing address is so large that it includes multiple colleges, check the five-digit zip code in the college’s mailing address. If the required data are available for this five-digit zip code, use these data. 3. If the city indicated in the mailing address is too small to have the required Census data, find the closest city for which data are available and use these data. Step 1 of the protocol is satisfied for the vast majority of colleges. We specified Step 2 to investigate further the case of colleges that, according to their mailing addresses, are in large cities served by multiple community colleges. Our hope was that by looking at a college’s fivedigit zip code, we could locate a smaller community within the larger metropolitan area for which required data are available. As it turned out, only for Los Angeles Southwest College were we successful in applying this step. For Los Angeles Southwest College, the city specified in the mailing address is Los Angeles, but the five-digit zip code indicates that the college is in Gardena. Since the required data are available for Gardena, we use employment data for Gardena rather than for Los Angeles. We were able to apply Step 3 to nine colleges whose local addresses indicate a small community in either a rural area or a large metropolitan 72 area. An example of the former is Cabrillo College in the small community of Aptos in Santa Cruz County. For this college, the closest larger city for which data are available is Santa Cruz. An example of the latter is Los Angeles Mission College in Sylmar. The required data are not available for Sylmar, but we could find data for nearby San Fernando. Overall, our protocol allows us to obtain unique employment data for 94 colleges. That is, we are forced to use duplicate data for 14 campuses. Los Angeles employment data are used for four Los Angeles area schools (out of nine Los Angeles Community College District campuses), and Oakland employment data are used for two Oakland area campuses (out of four campuses in the Peralta Community College District). Similarly, Sacramento employment data are used for all three colleges in the Los Rios Community College District serving Sacramento, San Diego employment data are used for all three colleges serving that city in the San Diego Community College District, and San Jose employment data are used for both colleges in the San Jose/ Evergreen Community College District. Local Service Area Demographics Our objective in specifying this set of variables is to measure for each college the demographic characteristics of its “local service area” defined in terms of prospective students, their parents, and local public officials who might be influential in determining curriculum mix. We considered several alternative approaches to defining the local service area. Initially, the approach that seemed the most promising was to make use of data measuring for each college the three-digit and five-digit origination zip code of all students enrolled in a community college as of Fall 2001. These data were kindly made available to us by Patrick Perry, Vice President for Management Information Systems in the Chancellor’s Office in Sacramento. Using 2000 Census data, the demographic information we require is also available at the three-digit and five-digit zip code levels, as well as for city and county. What we hoped was to find the zip codes (at the five-digit level or else at the three-digit level) that supplied a majority of a college’s students. Using this information to construct weights, we planned to calculate as a weighted average the 73 demographic characteristics of the local service area for each college. Unfortunately, we ran into the following problem. Enrollment at many colleges is so dispersed at the five-digit level that we were unable to isolate a reasonable number of five-digit zip codes that supply anywhere near a majority of student enrollment. On the other hand, the geographic areas covered by the relevant three-digit zip code were often so large that they failed to correspond to a reasonable definition of a local service area. Rather than trying to pin down the local service area in terms of three-digit and five-digit zip codes, the approach we ultimately followed was simply to make use of demographic data from the 2000 Census using information for the city in which the college is located. As indicated in Table 4.1, our community demographic variables include percentage of bachelor’s degree holders, percentage of foreign-born residents, median household income, and race and ethnicity. The city corresponding to each college, in turn, is specified using the protocol just outlined to measure local employer characteristics. 74 Appendix D Dataset for California Community Colleges Table D.1 Data Descriptions, Sources, and Descriptive Statistics Variable Var1 Var2 Var3 Var4 Var5 Var6 Var7 Var8 Var9 Var10 Var11 Var12 Var13 Var14 Var15 Var16 Var17 Var18 Var19 Var20 Description Attempted transfer, count enrollment Attempted basic skills, count enrollment Attempted nontransferable voc-ed, count enrollment Attempted all, count enrollments Attempted SAM Code A, count enrollment (apprenticeship) Attempted SAM Code B, count enrollment (advanced occupational) Attempted SAM Code C, count enrollment (clearly occupational) Attempted total voc-ed, count enrollment No. of colleges in district No. of SAM Code A courses No. of SAM Code B courses No. of SAM Code C courses No. of SAM Code D courses No. of SAM Code A programs No. of SAM Code B programs No. of SAM Code C programs No. of SAM Code D programs Total voc-ed programs Total “approved” voc-ed programs Total “approved” programs Data Source PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file VTEA VTEA VTEA VTEA VTEA VTEA VTEA VTEA VTEA Program inventory Program inventory Mean 56009 Min. Max. 2720 156094 4851 287 16819 5737 0 25573 74525 317 10128 184948 0 12667 3578 0 25734 12029 1360 46076 15924 1520 48866 2 19 17 0 195 148 4 994 403 29 1877 137 0 695 2 0 20 23 1 71 35 13 82 24 0 80 41 15 96 34 10 64 50 16 99 75 Table D.1 (continued) Variable Var21 Var22 Var23 Var24 Var25 Var26 Var27 Var28 Var29 Var30 Var31 Var32 Var33 Var34 Var35 Var36 Var37 Var38 Var39 Var40 Var41 Var42 Var43 Var44 Var45 Var46 Var47 Var48 Var49 Var50 Var51 Var52 Var53 Var54 Var55 Var56 Var57 Var58 Var59 Description Total student headcount, Fall 2001 Formulate career interest Voc-ed degree, no transfer Acquire job skills Update job skills Voc-ed certificate Maintain license Degree and transfer Transfer, no degree Degree, no transfer Basic skills Educational development Complete GED Undecided Uncollected Total students surveyed Community college emphasis (voc-ed = 1, transfer = 2, indeterminate = 3) Year college founded No. of male students, Fall 2001 No. of female students, Fall 2001 No. of students gender unknown No. of Asian students No. of black students No. of Filipino students No. of Hispanic students No. of Native American students No. of white students No. of nonresident alien students No response on race/ethnicity Other race/ethnicity No. of full-time students No. of part-time students Full-time/part-time unknown Community college district code Miles to nearest UC campus Miles to nearest CSU campus Minutes to nearest UC campus Minutes to nearest CSU campus No. of employees in manufacturing Data Source www.cpec.ca.gov FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 CPEC campus descriptions www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.ccco.edu www.yahoo.com www.yahoo.com www.yahoo.com www.yahoo.com 1997 Economic Census Mean 14377 89 55 149 88 72 25 626 204 102 64 94 32 404 75 2077 3 1949 6333 7960 84 1592 1042 469 3756 143 5942 225 932 276 3295 8781 2402 37 48 27 58 35 15666 Min. 1993 0 0 0 0 0 0 35 6 2 0 0 0 0 0 182 1 Max. 38412 299 451 943 988 517 112 2111 1912 1030 533 847 250 1771 1022 5102 3 1883 881 1111 0 27 13 7 77 16 165 0 1 0 279 661 175 1 1.7 1.5 2 3 0 2001 17580 21109 773 9955 5417 2590 19518 646 22859 2953 5277 1513 8625 23179 9703 72 282.5 178.2 334 170 186758 76 Table D.1 (continued) Variable Var60 Var61 Var62 Var63 Var64 Var65 Var66 Var67 Var68 Var69 Description No. of employees in wholesale trade No. of employees in retail trade No. of employees in real estate No. of employees in professional services No. of employees in administrative and support services No. of employees in educational services No. of employees in health care No. of employees in arts and entertainment No. of employees in accommodations and food services No. of employees in other services (except public administration) Data Source 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census Var70 Var71 Var72 Var73 Var74 Var75 Var76 Var77 Var78 Var79 Var80 Var81 Var82 % of population age 25+ with bachelor’s degree or higher % of population foreign born Median household income ($) Completion rate Transfer rate % white % black % Native American % Asian % Native Hawaiian and other Pacific Islander % some other race % two or more races % Hispanic ethnicity 2000 Census 2000 Census 2000 Census STRK STRK 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census Mean 7103 Min. Max. 10 87405 11377 10 118117 2775 10 39094 8978 10 122686 9479 25 104788 406 0 4599 6478 3 74565 1566 0 23403 9756 10 103746 2760 0 33761 26.5 5.4 78.1 23.3 49324 33.0 25.0 62.1 6.3 1.0 11.2 0.4 1.3 21900 11.5 0.0 16.7 0.2 0.1 0.3 0.1 54.4 173570 100.0 55.9 93.7 40.3 4.2 61.8 2.9 14.2 0.5 49.4 4.8 1.9 10.1 28.1 2.2 89.3 77 Table D.2 Construction of Curriculum Mix Variables from Raw Data Curriculum Variable Transfer credits/all credits Narrow voc-ed credits/all credits Basic skills credits/all credits Residual credits/all credits Broad voc-ed credits/all credits Narrow voc-ed credits/all voc-ed credits Apprenticeship credits/all voc-ed credits Advanced voc-ed credits/all voc-ed credits Academic transfer credits/all credits Apprenticeship courses/all courses Advanced voc-ed courses/all voc-ed courses Voc-ed courses/student Apprenticeship programs/student Advanced voc-ed programs/student Advanced courses/advanced programs Approved voc-ed programs/total programs Undecided student interest/all students Student interest in basic skills/all students Student interest in transferring/all students Student interest in voc-ed/all students Student interest in nontransfer programs/all students Transformation of Raw Data Var1/Var4 Var3/Var4 Var2/Var4 (Var4–Var1–Var2–Var3)/Var4 Var8/Var4 Var3/Var8 Var5/Var8 Var6/Var8 [Var1–(Var8-Var3)]/Var4 Var10/(Var10+Var11+Var12) Var11/(Var10+Var11+Var12) (Var10+Var11+Var12)/Var21 Var14/Var21 Var15/Var21 Var11/Var15 Var19/Var20 Var34/(Var36-Var35) (Var31+Var32+Var33)/(Var36-Var35) (Var28+Var29)/(Var36-Var35) (Var22+. . .+Var27)/(Var36-Var35) (Var22+. . .+Var27+Var30)/(Var36-Var35) 78 References Bailey, Thomas, “Community Colleges in the 21st Century: Challenges and Opportunities,” in Patricia A. Graham and Nevzer G. Stacey, eds., The Knowledge Economy and Postsecondary Education: Report of a Workshop, National Academy Press, Washington, D.C., 2002, pp. 59–75. Barnow, Burt S., and Jeffrey A. Smith, “Performance Management of U.S. Job Training Programs,” in Christopher J. O’Leary, Robert A. Straits, and Stephen A. Wandner, eds., Job Training Policy in the United States, W. E. Upjohn Institute for Employment Research, Kalamazoo, Michigan, 2004. California State Department of Education, Master Plan for Higher Education in California, 1960–1975, Sacramento, Calfornia, 1960. Chancellor’s Office, “Student Right-to-Know: Rates and Analysis for the 1999 Cohort,” Technology, Research, and Information Systems Division, California Community Colleges, Sacramento, California, June 20, 2003. Ehrenberg, Ronald G., and Christopher L. Smith, “Within State Transitions from 2-Year to 4-Year Public Institutions,” Economics of Education Review, Vol. 23, February 2004, pp. 11–28. Employment and Training Administration, “Common Measures Policy,” Training and Employment Guidance Letter No. 15-03, Washington, D.C., December 10, 2003. Grubb, W. Norton, Working in the Middle: Strengthening Education and Training for the Mid-Skilled Labor Force, Jossey-Bass, San Francisco, California, 1996. Gumport, Patricia H., “The Demand-Response Scenario: Perspectives of Community College Presidents,” Annals of the American Academy of Political and Social Science, Vol. 58, March 2003, pp. 38–57. 79 Johnson, Richard A., and Dean W. Wichern, Applied Multivariate Statistical Analysis, Prentice Hall, Englewood Cliffs, New Jersey, 1988. Joint Committee to Develop a Master Plan for Education, “The California Master Plan for Education,” Sacramento, California 2003. Kane, Thomas J., and Cecilia E. Rouse, “The Community College: Educating Students at the Margin Between College and Work,” Journal of Economic Perspectives, Vol. 13, Winter 1999, pp. 63–84. Laden, Berta V., “Two-Year Hispanic-Serving Colleges,” in Barbara Townsend, ed., Two-Year Colleges for Women and Minorities: Enabling Access to the Baccalaureate, Falmer Press, New York, 1999, pp. 151–194. Macro, Bronwen, Sherry Almandsmith, and Megan Hague, “Creating Partnerships for Workforce Investment: How Services Are Provided Under WIA,” revised final report, Berkeley Policy Associates, Oakland, California, September 2003. Romano, Richard M., “Financing Community Colleges Across the States: An Economic Perspective,” paper presented at the Cornell Higher Education Research Institute conference on “The Complex Community College,” Cornell University, Ithaca, New York, October 13–14, 2003. SAS Institute, SAS User’s Guide: Statistics, Version 5 edition, SAS Institute, Inc., Cary, North Carolina, 1990. State of California, “1996 Amendment to the California State Education Code: Chapter 1057,” Section 2, 1996. 80 About the Authors ANDREW M. GILL Andrew M. Gill is a professor of economics at California State University, Fullerton, where he specializes in labor economics and econometrics. His research explores community college issues, gender and racial discrimination, occupational wage differentials, and the labor market effects of drug use. He is currently co-editor of the journal Contemporary Economic Policy. He received his Ph.D. in 1985 from Washington State University. DUANE E. LEIGH Duane E. Leigh is a professor of economics at Washington State University, Pullman. Since the mid-1980s, his research has focused on the labor market effects of adult training and retraining programs designed for displaced workers and female welfare recipients. He received his Ph.D. in 1969 from Michigan State University. 81 Related PPIC Publications Financing California’s Community Colleges Patrick J. Murphy The Effects of School-to-Career Programs on Postsecondary Enrollment and Employment David Neumark School Finance and California’s Master Plan for Education Jon Sonstelie and Peter Richardson (editors), Julian R. Betts, Eric J. Brunner, Anne Danenberg, Jane Leber Herr, Susanna Loeb, Lawrence O. Picus, Peter Richardson, Heather Rose, Kim S. Rueben, and Jon Sonstelie (contributors) PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 83" } ["___content":protected]=> string(102) "

R 804AGR

" ["_permalink":protected]=> string(105) "https://www.ppic.org/publication/evaluating-academic-programs-in-californias-community-colleges/r_804agr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8280) ["ID"]=> int(8280) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:36:12" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(3433) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 804AGR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_804agr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_804AGR.pdf" ["wpmf_size"]=> string(6) "971557" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(164811) "Evaluating Academic Programs in California’s Community Colleges ••• Andrew M. Gill Duane E. Leigh 2004 Library of Congress Cataloging-in-Publication Data Gill, Andrew M. (Andrew Matthew) Evaluating academic programs in California’s community colleges / Andrew M. Gill, Duane E. Leigh p. cm. Includes bibliographical references. ISBN: 1-58213-100-7 1. California Community Colleges. 2. Community colleges— Curricula—California—Evaluation. 3. Community colleges— Curricula—California—Statistics. I. Leigh, Duane E. II. Public Policy Institute of California. III. Title. LB2361.6.C2G55 2004 378.1'543'09794—dc22 2004017225 Copyright © 2004 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. PPIC does not take or support positions on any ballot measure or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword California’s system of higher education is a three-legged stool. One leg of that stool—the California Community College (CCC) system— has the largest number of students. Sprawling over the state with 108 colleges in 72 districts, the CCC system enrolled well over 1.5 million students in academic year 2002. The other two legs of the stool—the University of California (UC) and California State University (CSU)— together enrolled less than 600,000 students in that same year. PPIC is committed to improving public understanding about the financing and operation of California’s community colleges because they are critically important in educating and training the state’s future labor force. In this study, Andrew Gill and Duane Leigh ask a key question about the function of the CCC system: To what extent do California’s community colleges choose different missions? This is an important question for two reasons. First, as part of the master plan for higher education, it is assumed that community colleges often represent a stepping stone to completion of a four-year education at one of the UC or CSU campuses. Do the community colleges perform this function? If not, alternative means of preparing students for four-year degrees will be required. Second, if community colleges are to be assessed for their contribution to higher education in the state, each campus will have to be judged against its own mission—not some generalized mission that seriously weakens the conclusions of a statewide evaluation of the system. In fact, the authors conclude that a “one-size-fits-all” strategy for community colleges may not be appropriate. Over 40 percent of the colleges specialize in one way or another. For example, Gill and Leigh find that most credits offered by most community colleges are transferable to four-year colleges. They note that both vocational education programs and traditional academic programs generate these credits. At the same time, many campuses provide vocational education iii courses that are not transferable but that are expected to lead to jobs in the local labor market. The authors estimate that 19 of California’s 108 community colleges emphasize nontransferable vocational education credits and de-emphasize transfer credits, whereas 26 community colleges do the opposite. The message of this report is clear: Although community colleges are part of a statewide system of higher education, their roots in local communities generate a greater variety of missions, curricula, and courses than is found in traditional four-year universities. This variety is to be expected, and presumably each campus serves the needs and requirements of its particular student body. It is likely that tensions will continue between the overall goals of the California system of higher education, which is funded in large part with state resources, and the goals of local community colleges, which rightly feel that they are more in touch with the preferences of families and employers in their respective cities and towns. This tension has probably been responsible, in part, for a declining share of total higher education spending allocated to community colleges. The tension is a reflection of those who see community colleges as an integral part of a three-legged stool and of others who prefer that community colleges exercise their freedom to break from the mold that constrains most college curriculum. There is no easy solution, but the authors of this report make it clear that the mission of each community college is important to its program of services and should therefore be judged accordingly. David W. Lyon President and CEO Public Policy Institute of California iv Summary Community colleges have traditionally received funding based on student enrollment, which is usually considered an input in the educational process. Recently, however, legislation enacted at the federal and state levels specifies that funding is to hinge, as least in part, on student performance—an output measure. Performance standards improve resource allocation by identifying colleges whose programs do not measure up to the standards so that remedial action can be taken or sanctions imposed. At the same time, performance standards applied uniformly across community colleges may be counterproductive if colleges differ significantly in their missions. Using data for community colleges in the California Community College System (CCCS), our study provides empirical evidence on the extent to which community colleges choose different missions. We ask three interrelated questions. First, do colleges differ in their missions as reflected in the mix of academic programs they offer? The size and complexity of the CCCS make it likely that individual colleges will choose different missions, which we measure as differences in curriculum “emphasis” and “specialization.” Differences in curriculum emphasis are defined by systematic deviations in the curriculums offered by individual colleges from those offered by a “typical” or average college. Curriculum specialization goes a step further by linking an emphasis on one curriculum measure with a de-emphasis on another. If colleges do appear to differ in their missions, our second question is whether observed differences in emphasis and specialization can be explained using measurable indicators of college-specific characteristics and community needs. If community colleges differ in ways that are linked to these explanatory variables, the third question is whether our evidence may be useful in developing guidelines for evaluating community college performance. v Evidence on the missions of community colleges is typically obtained by interviewing college administrators and stakeholders in the community. The approach we pursue in this report, in contrast, is to collect in one large dataset curriculum information available online for all 108 CCCS campuses. One element of this dataset is our characterization of whether individual colleges have different curriculum emphases as indicated by their campus descriptions. In addition, we collect a wealth of quantitative data on curriculum mix measured in terms of credits, courses, programs, and the interests of freshman students. To this curriculum information, we append data available online measuring local labor market characteristics, community demographic variables, and college-specific characteristics such as membership in a multicampus community college district and proximity to a four-year college. In response to our first research question, we find that most community colleges are oriented toward offering programs that supply credits transferable to four-year colleges. We note, however, that transferable programs include both traditional academic programs and vocational education (voc-ed) programs such as programs in business, computer science, and electronics. Transferable voc-ed offerings are increasingly common in California and nationwide. Where California community colleges do seem to differ is in their emphasis on voc-ed offerings that are transferable and taught at an advanced level. For example, the interquartile range calculated for a curriculum variable measuring the proportion of transferable voc-ed credits indicates that the top 25 percent of colleges differ from the bottom 25 percent by at least 27 percentage points. Similarly, the top 25 percent of colleges differ from the bottom 25 percent by nearly 20 percentage points on the proportion of voc-ed courses taught at an advanced level. Moving beyond emphasis to the concept of specialization, our preferred measure suggests that 45 colleges specialize in some way. Specifically, 26 colleges have a transfer specialization, meaning that they choose to emphasize transfer credits while de-emphasizing nontransferable voc-ed credits. Another 19 colleges specialize in nontransferable voc-ed, meaning that they emphasize nontransferable voc-ed credits while de-emphasizing transfer credits. vi Turning to the second research question, the intercollege differences in curriculum emphasis and specialization we observe are reasonably well explained by measurable indicators of college-specific characteristics and community needs. Colleges that specialize in either a transferable curriculum or in nontransferable voc-ed offerings are more likely than other colleges to belong to a multicampus district and to be located in large labor markets. Between these two categories of specialization, colleges that have a transfer specialization tend to be closer to a UC or a CSU campus. Colleges that specialize in nontransferable voc-ed tend to have a larger proportion of Hispanic students and to be in larger communities with greater minority populations. Building on our empirical evidence, we offer the following observations that we believe should be considered in evaluating the performance of California’s community colleges. 1. On average, California community colleges offer more credits in transferable programs than in other broad curriculum categories. With a few exceptions, our evidence suggests that colleges are heavily engaged in offering transferable curriculums. 2. Although colleges place a priority on transferable offerings, attention to transfer rates is not sufficient for evaluation because transferable curriculums are a mix of voc-ed courses and traditional academic courses. Success for colleges with a strong transferable voc-ed orientation might be placement in trainingrelated jobs as opposed to transfers to four-year colleges. 3. Some colleges specialize in transferable curriculums, whereas others specialize in nontransferable voc-ed. Similarly, colleges differ in their emphasis on advanced voc-ed curriculums and in the interests of their freshman students in voc-ed. These differences in specializations and emphases indicate important differences in the missions chosen by different colleges. Differences in missions, in turn, suggest that a “one size fits all” evaluation strategy is not appropriate. 4. Basic skills programs are a small proportion of most community colleges’ total credits offered. vii 5. Differences in missions as reflected in differences in curriculum specializations and emphases are responsive to differences in college-specific characteristics and community needs. Precisely how to allow for differences in missions in evaluating performance is a difficult issue that is currently receiving attention by policymakers in states across the nation. We conclude our report by describing two possible approaches. The first is a model-based approach that allows common performance standards to be adjusted in response to quantitative differences between colleges in student characteristics and local economic conditions. To the extent that colleges can demonstrate that their missions differ, the second approach is for the state to tailor performance standards to take into account alternative missions. viii Contents Foreword....................................... iii Summary........................................v Figures ....................................... xiii Tables ......................................... xv Acknowledgments................................ xvii 1. INTRODUCTION .............................1 Multiple Missions and Reporting Requirements..........3 Research Objectives .............................6 Organization of the Report ........................7 2. MEASURES OF COMMUNITY COLLEGE CURRICULUM MIX ...........................9 CPEC Campus Descriptions ......................10 PFE Data on Enrollment.........................12 Goal 3 Measures of Broad Curriculum Categories......13 Goal 4 Variables Measuring Categories of Voc-Ed Credits ..................................15 Breaking Out Traditional Academic Transfer Credits ..................................16 VTEA Data Measuring Courses and Programs and Program Inventory Data .....................16 Voc-Ed Course Measures .......................17 Voc-Ed Program Measures ......................18 First-Time Freshmen Data........................19 Summary....................................22 3. CURRICULUM SPECIALIZATION ...............23 Getting Started with Histograms ...................24 What Do IQ Ranges Tell Us About Differences in Emphasis?................................25 Evidence of Specialization ........................29 ix Correlation Result 1: Transferable Curriculum vs. Nontransferable Voc-Ed......................29 Correlation Result 2: Transferable Curriculum vs. Proportion of Voc-Ed That Is Nontransferable ......35 Correlation Result 3: Positive Relationships Between Measures of Advanced Voc-Ed Curriculum ........36 Correlation Result 4: Tradeoff Between Student Interests in Voc-Ed and Transferring.............36 Bringing It Together Using Factor Analysis ............37 Summary....................................40 4. EXPLAINING DIFFERENCES IN CURRICULUM MIX .......................................41 Description of Explanatory Variables ................41 College-Specific Characteristics...................41 Local Job Opportunities........................44 Community Demographics .....................45 Explaining Differences in Specializations and Emphases ................................46 Specializations in Transferable Curriculums and Nontransferable Voc-Ed......................46 Emphasis on Advanced Voc-Ed ..................49 Student Interest in Voc-Ed......................51 Summary....................................52 5. POLICY DISCUSSION .........................53 Key Provisions of the California Master Plan ...........53 Performance Evaluation Requirements ...............54 Guidance to Policymakers for Evaluating Community Colleges .................................56 What does Our Evidence Show? ..................57 Allowing for Differences in Missions ...............60 Appendix A. Technical Detail on Measures of Curriculum Mix .......63 B. Technical Detail on Factor Analysis .................67 C. Data Sources and Technical Details Used to Construct the Explanatory Variables ...........................71 D. Dataset for California Community Colleges ...........75 x References ......................................79 About the Authors ................................81 Related PPIC Publications...........................83 xi Figures 3.1. Ratio of Transferable Credits to All Credits .........24 3.2. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits ............................25 3.3. Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Credits ..................................31 3.4. Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits ............................36 xiii Tables 1.1. Percentage Completion and Transfer Rates for Full-Time Freshmen Students Enrolled in California Community Colleges, Fall 1999 to Spring 2002.......4 2.1. California Community Colleges’ Emphases Based on CPEC Campus Descriptions ...................12 2.2. Measures of Curriculum Mix Using PFE Enrollment Data, 2000–2001 ...........................14 2.3. Measures of Curriculum Mix Using VTEA Course and Program Data and Program Inventory Data, 1998–1999 ...............................17 2.4. Measures of Curriculum Mix Using FTF Student Academic Objectives, 1997 ....................20 3.1. Variability in Curriculum Mix Measures Using Data for Credits, Courses, Programs, and Student Objectives ................................26 3.2. Estimated Correlation Coefficients Between Selected Curriculum Mix Measures.....................30 3.3. Colleges Emphasizing and Specializing in a Transferable Curriculum, Alternative Definitions of Specialization ............................33 3.4. Colleges Emphasizing and Specializing in Nontransferable Voc-Ed, Alternative Definitions of Specialization ............................34 3.5. Top Five and Bottom Five Colleges Ordered by Factor Scores for Factors 1–4 ...................39 4.1. Descriptive Statistics for Explanatory Variables ......42 4.2. Cross-Tabulations Between Specializations in Transferable Curriculums and Nontransferable Voc-Ed and Selected Explanatory Variables .........47 4.3. Cross-Tabulations Between Emphasis on Advanced Voc-Ed Courses and Student Interest in Voc-Ed and Selected Explanatory Variables ..................50 xv B.1. Factor Loadings from the Orthogonal Rotated Factor Patterns .............................68 B.2. Descriptive Statistics for Factor Scores Obtained from Factor Analysis .........................70 D.1. Data Descriptions, Sources, and Descriptive Statistics .................................75 D.2. Construction of Curriculum Mix Variables from Raw Data.................................78 xvi Acknowledgments The financial support of the Public Policy Institute of California is gratefully acknowledged. Lorana Chaparro, Stephanie Lomibas, and Juan Mansilla provided outstanding research assistance. Helpful comments on the first draft of this report were received from Norton Grubb, Christopher Jepsen, Paul Lewis, David Mertes, Fred Silva, and Jon Sonstelie. Joyce Peterson and Peter Richardson provided excellent editorial assistance in rewriting subsequent drafts. Ron Mittelhammer got us off to a good start on our application of factor analysis. We are also grateful for the helpful suggestions of participants at a PPIC seminar held in October 2003. xvii 1. Introduction Community colleges are complex institutions offering a wide range of academic programs to various groups. Historically, community college students completed two years of a general undergraduate education and earned an associate’s degree (A.A.). Students with the interest and capability would then transfer to a four-year college to complete a bachelor’s degree. Kane and Rouse (1999) describe how over time community colleges have also expanded their course offerings in vocational education (voc-ed) and adult basic skills. Voc-ed programs often culminate in an A.A. degree or certificate. But Grubb (1996, p. 94) emphasizes that today, credits earned in such voc-ed programs as business, computer science, and electronics are frequently transferable to four-year colleges. In addition to offering a broad range of courses in traditional academic areas, voc-ed, and adult basic skills, a growing number of community colleges provide “contract training”—that is, occupational skills classes designed to meet the needs of particular employers and sometimes delivered on site. As academic offerings have expanded, so has the number of groups to which community colleges must be responsive. Traditional students are 18–22-year-olds who attend community college full-time with the objective of transferring to a four-year college. Nontraditional community college students include adult workers returning to school to sharpen their skills or earn a degree, dislocated workers and returning homemakers seeking retraining for new careers, single mothers seeking to make the welfare-to-work transition, and immigrants interested in improving their language skills and learning more about American culture. These nontraditional students are typically enrolled part time. Other community college constituencies include employers looking for contract training programs and local government officials seeking to encourage the economic development of their communities. 1 As their program offerings and constituencies expand, community colleges are increasingly being held to performance standards. The primary objective of performance standards—improving resource allocation—is laudable. These standards help colleges identify weak programs and either improve or drop them from the curriculum. They also help community college districts and state systems, such as the California Community College System (CCCS), identify underperforming colleges in need of remedial action or sanctions. The downside of performance standards is that community colleges may not be all alike. When colleges differ substantially in their missions, performance standards applied uniformly will penalize colleges that are accomplishing the mission they have chosen. For example, transfer rate is not an appropriate performance criterion for a community college whose primary mission is to provide occupational training for immediate employment in the local labor market. In this report, we investigate the extent to which California community colleges differ in their missions, which we measure by differences in the mix of academic programs they offer. We focus on California community colleges for four reasons. First, the CCCS is the nation’s oldest and most extensive community college system offering educational services on 108 campuses to over 1.6 million students as of Fall 2001. Romano (2003) remarks that California so dominates national community college statistics that researchers often do their calculations with and without California data included. In a community college system as large and complex as California’s, community colleges are most likely to differ in their missions.1 Second, California is well known for and has been widely emulated by other states for its open admission and low tuition policies. These policies were established in 1960 by the Master Plan for Higher Education in California. Third, the rich ethnic and racial diversity of the state allows us to address issues involving the effect of race and ethnicity on the mix of community _____________ 1As Bailey (2002, p. 69) notes, community colleges cannot be expected to do everything well and therefore must choose a more limited set of objectives in much the same way that a business firm chooses to focus on its “core competency.” At the same time, he indicates that, largely for political reasons, community colleges nationwide often pursue a comprehensive strategy emphasizing breadth of program offerings. 2 college academic offerings. Finally, a wealth of data is available online identifying the academic programs provided by California community colleges. To our knowledge, our report is the first to bring together the several sources of curriculum data for California community colleges into a unified and consistent dataset. Multiple Missions and Reporting Requirements The Master Plan of 1960 gives California community colleges considerable discretion in determining the mix of academic programs to offer. At the same time, California community colleges, as is true for community colleges across the nation, face a plethora of performance standards. One example is the federal student-right-to-know (SRTK) regulation that requires the collection and reporting of transfer and completion rates. To set the stage for our discussion of differences in missions among colleges, Table 1.1 displays transfer and completion rates for 107 colleges in the CCCS measured as of 1999 (Chancellor’s Office, 2003). Several measurement issues make us cautious in comparing the performance of particular colleges using SRTK data.2 The main point we draw from Table 1.1 is simply the variability across colleges in the two rates shown. Completion rates range from a high of 49.5 percent for Las Positas College to a low of 11.5 percent for Los Angeles Mission College. Similarly, transfer rates range from a high of 55.9 percent for Merritt College to a low of 4.6 percent for Lassen College. If we take these data at face value, a question that comes to mind is, what are colleges with low completion and transfer rates doing? One answer is that they are not doing their job and need to improve. An equally plausible answer, and one that is noted in the Chancellor’s Office (2003) report, is that colleges differ in their missions. As suggested above, some colleges may choose not to emphasize the transfer function but instead _____________ 2The first of these is the restriction of the universe of students to first-time, fulltime freshmen (which greatly reduces the tracking pool). The second is a counting methodology that requires a hierarchy of outcomes such that students who earned degrees or became transfer-prepared are removed from the calculation of transfer rates. The third issue is inclusion in transfer rates of students who transfer between community colleges in addition to students who transfer to four-year institutions. 3 Table 1.1 Percentage Completion and Transfer Rates for Full-Time Freshman Students Enrolled in California Community Colleges, Fall 1999 to Spring 2002 4 College Alameda Allan Hancock American River Antelope Valley Bakersfield Barstow Butte Cabrillo Canada Canyons Cerritos Cerro Coso Chabot Chaffey Citrus Coastline Columbia Compton Contra Costa Cosumnes River Crafton Hills Cuesta Cuyamaca Cypress DeAnza Desert Diablo Valley East Los Angeles El Camino Completion Rate 33.7 41.9 33.6 43.3 34.6 29.6 36.8 35.1 23.1 40.7 27.1 39.8 33.8 25.7 28.3 26.7 32.1 17.4 32.5 27.6 29.9 35.3 28.1 20.8 42.8 48.4 41.1 26.1 38.6 Transfer Rate 38.4 11.0 13.4 10.1 23.3 12.7 17.2 17.1 41.0 23.5 20.8 18.3 30.5 18.9 26.8 33.3 30.4 21.7 22.5 25.5 20.5 25.7 28.1 41.2 33.4 4.8 24.3 22.5 16.1 College Evergreen Valley Feather River Foothill Fresno City Fullerton Gavilan Glendale Golden West Grossmont Hartnell Imperial Valley Irvine Valley Lake Tahoe Laney Las Positas Lassen Long Beach Los Angeles City Los Angeles Harbor Los Angeles Mission Los Angeles Pierce Los Angeles Southwest Los Angeles Trade-Tech Los Angeles Valley Los Medanos Marin Mendocino Merced Merritt Completion Rate 28.4 48.5 39.2 30.2 23.7 37.9 32.8 35.5 30.2 44.1 46.4 39.7 35.0 28.0 49.5 23.1 30.2 29.4 29.8 11.5 36.7 23.1 21.6 38.3 24.3 26.5 34.4 30.8 23.7 Transfer Rate 40.1 33.3 31.5 32.8 47.7 22.7 25.5 29.3 24.2 15.3 16.6 23.7 52.5 37.9 22.4 4.6 18.6 25.2 22.8 26.9 30.7 15.4 35.1 28.3 29.2 36.7 23.2 15.2 55.9 Table 1.1 (continued) 5 College Mira Costa Mission Modesto Monterey Moorpark Mt. San Antonio Mt. San Jacinto Napa Valley Ohlone Orange Coast Oxnard Palo Verde Palomar Pasadena City Porterville Redwoods Reedly Rio Hondo Riverside Sacramento City Saddleback San Bernardino San Diego City San Diego Mesa San Diego Miramar Completion Rate 27.8 32.0 34.6 41.5 29.9 31.6 29.8 35.0 33.8 40.3 27.4 31.8 28.2 39.7 41.6 31.7 41.0 23.7 32.1 33.1 38.2 15.7 13.1 28.4 17.1 Transfer Rate 25.9 33.3 18.0 18.7 27.3 28.5 20.2 10.0 26.8 26.7 19.2 18.2 21.0 20.2 13.6 23.2 38.5 32.2 27.2 23.9 22.1 20.4 21.3 30.5 51.4 College San Francisco City San Joaquin Delta San Jose City San Mateo Santa Ana Santa Barbara Santa Monica Santa Rosa Santiago Canyon Sequoias Shasta Sierra Siskiyous Skyline Solano Southwestern Taft Ventura Victor Valley Vista West Hills West Los Angeles West Valley Yuba Completion Rate 43.5 32.4 17.9 41.7 36.3 43.8 36.8 45.7 31.1 34.2 34.7 36.0 30.3 31.9 29.6 36.5 42.0 32.8 23.4 22.1 23.3 17.4 40.6 38.1 Transfer Rate 20.1 17.3 38.4 33.1 25.0 26.6 19.1 9.7 37.8 15.3 16.7 22.0 25.8 24.8 24.4 11.6 14.5 26.6 22.3 48.8 17.0 32.6 31.5 25.3 SOURCE: Chancellor’s Office (2003). NOTES: Completion rates measure the percentage of students who earn an A.A. degree or certificate or become “transfer-prepared.” Excluding those measured as “completers,” transfer rates measure the percentage of students who transfer to a four-year institution or to another two-year college. Copper Mountain College is omitted, since it did not begin operation until 2001. choose to emphasize voc-ed training that provides occupational skills immediately salable in the local labor market. Before leaving Table 1.1, it is worth noting that the dataset used to calculate these rates—the FirstTime Freshman (FTF) student cohort—is discussed in Chapter 2 as a measure of the educational objectives of freshman students. In addition to SRTK regulation, community colleges are subject to reporting requirements imposed by two important pieces of federal legislation. These are the Workforce Investment Act (WIA) of 1998 and the Carl D. Perkins Vocational and Technical Education Act (VTEA) of 1998. These laws impose requirements that are substantially more detailed than those needed to satisfy colleges’ SRTK obligations. In particular, both laws require that states establish comprehensive accountability systems that include multiple “core indicators” of performance. In California, the Partnership for Excellence (PFE) agreement of 1996 between the state and the CCCS and the California Perkins State Plan of 1994 specify performance standards that satisfy federal requirements. With some justification, community college administrators often regard these performance standards as flawed. For our purposes, however, a favorable but unintended consequence of performance standards is that California community colleges report not only outcome measures, such as transfer and completion rates, but also measures of the academic programs they provide. As will be discussed in more detail in Chapter 2, these curriculum measures are expressed in terms of distributions of credits, courses, programs, and student interests. Existing analyses of the question “What do community colleges do?” typically proceed with site visits to selected colleges. Our investigation of a dataset covering all the community colleges in a large system such as the CCCS should allow us to provide a more comprehensive answer to this fundamental question. Research Objectives Our analysis has three primary goals. The first is to measure the extent to which community colleges differ in their missions as measured by the curriculum mixes they offer. We discuss this variation in terms of differences in curriculum “emphasis” and curriculum “specialization.” In 6 Chapter 3, we consider alternative definitions of these two concepts. By differences in curriculum emphasis, we mean systematic deviations in the curriculums offered by individual colleges from that offered by a typical or average college. By differences in curriculum specialization, we go a step further by coupling emphasis on one curriculum variable (for example, traditional academic programs leading to transfer to a four-year college) with a de-emphasis on another variable, such as occupational skills training immediately applicable in the local labor market. The second goal is to explain observed differences in emphasis and specialization using measures of college-specific variables, such as proximity to the nearest four-year college, and of community needs, such as local job opportunities and demographic characteristics of the local service area. Our final goal is to use this evidence to help guide policymakers in evaluating community colleges. Organization of the Report Chapter 2 outlines the several sources of online information we put together into a consistent dataset. This dataset includes, for each of the 108 campuses in the CCCS, a number of alternative curriculum measures. These measures receive detailed examination in Chapter 3, where we look for differences in curriculum emphasis and specialization across colleges. In Chapter 4, we present evidence on the extent to which differences in curriculum emphases and specializations can be accounted for by a reasonable set of explanatory variables including measures of community demographics, local labor market characteristics, and college-specific characteristics. In Chapter 5, we offer an overview of key provisions of the California Master Plan and provide some detail on the multiple performance indicators faced by community colleges. Then we draw on our empirical findings to offer suggestions to policymakers intended to better inform the process of evaluating the performance of community colleges. 7 2. Measures of Community College Curriculum Mix In Chapter 1, we identified the three primary missions of colleges in the California Community College System. This chapter is concerned with measuring these three academic missions using data available online for the 108 CCCS campuses. In this chapter, we present simple descriptive statistics for the curriculum measures developed. A more detailed analysis of these data looking for evidence of curriculum emphasis and specialization is in Chapter 3. One approach to measuring a community college’s primary mission or missions is to examine how the college describes itself on its web page. We begin this chapter with a consideration of these campus descriptions, which are collected on the California Postsecondary Education Commission (CPEC) web site. A second approach to measuring differences in missions is to use online sources of quantitative data measuring curriculum mix. As noted in Chapter 1, California community colleges report a wealth of curriculum data in response to performance standards imposed at the state and federal levels. In this chapter, we describe three sources of quantitative data from which we construct 21 curriculum mix variables. Sources of quantitative data are • PFE data on student enrollment measured in credits; • Data on voc-ed courses and programs reported by the CCCS Chancellor’s Office to satisfy its VTEA obligation; and • FTF data on students’ academic objectives. Added to our discussion of VTEA data is a brief description of the Chancellor’s Office “Inventory of Approved and Projected Programs,” which provides an additional measure of “approved” voc-ed programs. Technical detail on these quantitative datasets is found in Appendix A. 9 CPEC Campus Descriptions For each community college, the CPEC web site provides a brief campus description, information on total enrollment, the gender and race or ethnicity of students, a list of programs offered, and links to additional information about the college. Also reported is a campus mailing address. The campus descriptions provide information indicating what the college believes to be especially noteworthy about its academic programs. Using this information to measure curriculum emphasis, we construct a three-category discrete variable with categories indicating voc-ed emphasis, transfer emphasis, and emphasis indeterminate. The criteria used to classify each college in this threecategory scheme are as follows. Voc-ed emphasis • Pride expressed in particular occupational skill training programs; • Emphasis on partnerships with local business firms; and • Emphasis on employment opportunities in the local area and job placement assistance services. Transfer emphasis • Expression of pride in college’s success in transferring students to University of California (UC) and California State University (CSU) campuses; and • Emphasis on special transfer arrangements with four-year colleges, especially UC and CSU campuses. Indeterminate • Emphasis on breadth of course offerings, student services, and extracurricular programs; • Emphasis on location and facilities; and • Emphasis on other attributes such as multicultural student body, small classes, individual attention, and high-quality faculty. 10 Two colleges that are easily classified as having a voc-ed or transfer emphasis are, respectively, Los Angeles Trade-Tech College and Santa Barbara City College (SBCC). Los Angeles Trade-Tech describes itself as “the ‘flagship vocational college’ of the Los Angeles Community College District.” On the other hand, SBCC writes that “the college prides itself on the overall excellence of its academic programs and student support services. It is a top community college for transferring students to the University of California system.” The SBCC statement goes on to describe the college’s special transfer programs with area four-year colleges and universities. Laden (1999) identifies SBCC and Santa Monica College as the state’s leading community colleges in terms of overall transfer rates and transfers to the highly selective UC system. In the discussion of PFE, VTEA, and FTF data that follows, we use Los Angeles Trade-Tech and Santa Barbara City College as examples to illustrate the variety of curriculum mix measures provided by these datasets. Although Los Angeles Trade-Tech and SBCC neatly fit the criteria outlined above, many other colleges provide campus descriptions that are either too brief or too general to allow ready classification. In these cases, we followed up by examining the college’s mission statement, the president’s introductory letter if available, and other pertinent information. This additional information was seldom helpful. As seen in Table 2.1, we are able to classify 17 colleges, including Los Angeles Trade-Tech, as having a voc-ed emphasis, and 11 colleges, including Santa Barbara City College, as having a transfer emphasis. The remaining 80 colleges fall into the indeterminate emphasis category. Our general impression from reading CPEC descriptions is that most colleges do not take advantage of this opportunity to differentiate themselves when describing their academic programs. Rather, they indicate that they provide a full range of educational programs and attempt to distinguish themselves by emphasizing their location and facilities. 11 Table 2.1 California Community Colleges’ Emphases Based on CPEC Campus Descriptions College with Transfer Emphasis Chabot Diablo Valley Fullerton Las Positas Los Angeles Harbor Los Angeles Pierce Marin Oxnard Santa Barbara City Santa Monica Ventura College with Voc-Ed Emphasis Coastline Contra Costa Copper Mountain Cosumnes River Desert East Los Angeles Foothill Fresno City Lassen Long Beach City Los Angeles Trade-Tech Moorpark San Diego Miramar San Joaquin Delta Shasta Siskiyous Taft PFE Data on Enrollment The Partnership for Excellence is an agreement between the state and the CCCS committing community colleges to five quantitative performance goals, two of which are interesting to us. Goal 3 is the system’s commitment to increase overall rates of successful course completion, where courses are broken down by transferable, voc-ed, and basic skills credits. Goal 4 involves a CCCS commitment to enhance statewide workforce development by increasing successful course completion among voc-ed courses classified as apprenticeship, advanced occupational, and introductory occupational. Advanced occupational courses must have an introductory prerequisite in the same program area. Examples of advanced voc-ed courses are dental pathology, legal secretarial procedures, contact lens laboratory, fire hydraulics, and real estate finance. Introductory occupational courses are intended to provide students with entry-level job skills. Some examples of 12 introductory voc-ed courses are principles of advertising, clinical techniques, and technical engineering. Goal 3 Measures of Broad Curriculum Categories PFE data provided in response to Goal 3 are especially valuable because these data measure the quantitative importance of all three primary community college missions. Transferable courses are defined as courses that are transferable to UC or CSU campuses. Voc-ed courses, in contrast, are defined as nontransferable courses implying a “narrow” definition of voc-ed. Basic skills courses include those defined as “possibly occupational” and “nonvocational.” It should be emphasized that the transfer, narrow voc-ed, and basic skills categories are mutually exclusive. We use credits for courses attempted (as opposed to credits for courses completed) as our PFE curriculum mix measure because credits for courses attempted represent the academic services a college offers. One further point to note about Goal 3 PFE data is that the sum of transferable, narrow voc-ed, and basic skills credits attempted does not sum up to total credits attempted. We define as a “residual” category of course enrollment the difference between total credits attempted and the sum of enrollment in transferable, narrow voc-ed, and basic skills courses. Residual courses seem to us to fall under the heading of “consumption goods,” defined as courses in which students choose to enroll to enhance the quality of their lives as opposed to improving their labor market opportunities or building up transferable credits. The first four rows of Table 2.2 display means and maximum and minimum values for ratios of transferable, narrow voc-ed, basic skills, and residual credits to all credits. Also shown are the ratios calculated for SBCC and Los Angeles Trade-Tech. Row 1 indicates that on average, about 73 percent of credits offered during the 2000–2001 academic year are transferable. The ratio calculated for SBCC lies somewhat above this mean, whereas the ratio for Los Angeles Trade-Tech falls somewhat below. A maximum value of 92.2 percent is found for Orange Coast College in Orange County, and the minimum of 18.7 percent is calculated for Taft College at the southern end of the San Joaquin Valley. It is interesting to note from Table 2.1 that we failed to classify Orange Coast College as having a transfer emphasis because of its brief 13 Table 2.2 Measures of Curriculum Mix Using PFE Enrollment Data, 2000–2001 Enrollment Measure 1. Transferable credits/all credits 2. Narrow voc-ed credits/all credits 3. Basic skills credits/all credits 4. Residual credits/all credits 5. Broad voc-ed credits/ all credits 6. Narrow voc-ed credits/ all voc-ed credits 7. Apprenticeship credits/ all voc-ed credits 8. Advanced voc-ed credits/ all voc-ed credits 9. Academic transfer credits/ all credits SBCC .854 Los Angeles TradeTech .624 Mean .733 Minimum Maximum .187 .922 .058 .162 .087 .000 .756 .041 .093 .069 .008 .215 .047 .121 .111 .015 .326 .254 .252 .229 .063 .821 .227 .643 .356 .000 .920 .000 .057 .017 .000 .727 .614 .180 .207 .000 .647 .658 .534 .591 .117 .859 and nonspecific CPEC campus description. However, we did classify Taft College as having a voc-ed emphasis on the basis of its CPEC description. The remaining three categories of credits shown in rows 2–4 are, on average, roughly evenly distributed across the narrow voc-ed, basic skills, and residual categories. In each of these three rows, in addition, ratios calculated for Los Angeles Trade-Tech are larger than those for SBCC. A large range of 75.6 percentage points between maximum and minimum values appears in row 2 for narrow voc-ed. The maximum value for this ratio is calculated for Taft College, and the minimum value is for Orange Coast College. Maximum/minimum ranges are much smaller for basic skills credits (20.7 percentage points) and residual credits (31.1 percentage points). The mean for proportion of basic skills credits (6.9%) in Table 2.2 probably understates the resources that 14 colleges devote to basic skills, since basic skills programs are typically noncredit programs (Gumport, 2003). Goal 4 Variables Measuring Categories of Voc-Ed Credits Whereas Goal 3 PFE data offer a more complete view of curriculum mix, Goal 4 PFE data provide a detailed breakdown of voc-ed credits by transferability and type. As indicated in Chapter 1, Grubb (1996) emphasizes that a growing proportion of voc-ed enrollment is in courses that are transferable to four-year colleges. He notes, in particular, that courses in business, computer science, and electronics are frequently transferable. A comparison of rows 2 and 5 of Table 2.2 indicates that, measured at the means, the proportion of voc-ed credits to all credits nearly triples to 22.9 percent when we consider the broad rather than the narrow definition. In row 5, inclusion of transferable voc-ed makes the broad voc-ed ratios for SBCC and Los Angeles Trade-Tech essentially the same at about 25 percent, whereas row 2 indicates a much lower narrow voc-ed ratio for SBCC. This difference in mix of voc-ed credits is highlighted in row 6 in the ratio of narrow voc-ed credits to all voc-ed credits. Almost two-thirds of voc-ed credits measured for Los Angeles Trade-Tech are nontransferable, as opposed to less than 25 percent for Santa Barbara City College. Rows 7 and 8 make use of Goal 4 PFE information on credits broken down by type of occupational education courses. Row 7 shows that SBCC has no enrollment in apprenticeship courses, whereas only 5.7 percent of total voc-ed enrollment at Los Angeles Trade-Tech is in apprenticeship courses. Well over half of all colleges resemble SBCC in having no students enrolled in apprenticeship programs. Nevertheless, apprenticeship training does represent a sizable component of voc-ed enrollment at a handful of colleges. The leading example is Santiago Canyon College, in Orange in Southern California, in which 72.7 percent of voc-ed enrollment is in apprenticeship courses. Santiago Canyon is clearly an outlier, however, because the next highest ratio of apprenticeship programs is just 12.3 percent for American River College. Row 6 shows that a much larger proportion of voc-ed credits offered at Santa Barbara City College is transferable than is the case at Los Angeles Trade-Tech. Row 8 indicates, similarly, that many more credits 15 are taught at the advanced level at SBCC than at Los Angeles TradeTech. The minimum level of 0 percent observed in row 8 is for Lake Tahoe Community College, followed by Taft College with only 2.8 percent of voc-ed enrollment at the advanced level. The maximum value of 64.7 percent is observed for Rio Hondo College. Breaking Out Traditional Academic Transfer Credits Knowing both nontransferable voc-ed credits and total voc-ed credits, we can readily calculate transferable voc-ed credits, which, when subtracted from total transfer credits, yields a measure of traditional academic transfer credits. Examples of traditional academic transfer courses are introduction to biology and introduction to sociology. The ratio of academic transfer credits to all credits is shown in line 9 of Table 2.2. Measured at the means in rows 1 and 9, academic transfer credits make up on average about 81 percent (.591/.733) of all transferable credits. The ratio of academic transfer credits to all credits calculated for Santa Barbara City College in row 9 exceeds that for Los Angeles TradeTech by 12.4 percentage points. The even larger difference of 23 percentage points observed in row 1 is clearly due to the greater emphasis of SBCC on transferable voc-ed courses. Minimum and maximum values of the academic transfer credit ratio are observed for Los Angeles Mission College (11.7 percent) and Orange Coast College (85.9 percent), respectively. VTEA Data Measuring Courses and Programs and Program Inventory Data We view the PFE enrollment data summarized in Table 2.2 as approximating an “equilibrium” description of a college’s curriculum mix, where the term equilibrium is used to mean a balance between current student demand and the college’s supply of courses and programs. Recognizing that current offerings of courses and programs hinge in part on past student demand, we view VTEA data as primarily capturing the current supply of courses and programs. From this perspective, VTEA data give us an alternative and independent measure of voc-ed curriculums. 16 Voc-Ed Course Measures Beginning with VTEA course data, rows 1 and 2 of Table 2.3 display ratios of apprenticeship and advanced occupational courses to total voc-ed courses, respectively. The maximum value for the apprenticeship course ratio in row 1 is obtained for Santiago Canyon College. Recall from the discussion of Table 2.2 that Santiago Canyon was singled out as an outlier for our measure of apprenticeship credits to total voc-ed credits. Santa Barbara City College is seen in the second row to have a much higher ratio of advanced occupational courses to all voc-ed courses than is the case for Los Angeles Trade-Tech. The maximum value of this ratio (83.3 percent) is obtained for Mt. San Jacinto College. Note that the sample size in Table 2.3 is 107 rather than 108 campuses because data for Copper Mountain College are not available. Since VTEA data do not provide information on total courses, we were unable to calculate a ratio of voc-ed courses to all courses Table 2.3 Measures of Curriculum Mix Using VTEA Course and Program Data and Program Inventory Data, 1998–1999 Course or Program Measure 1. Apprenticeship courses/ all voc-ed courses 2. Advanced voc-ed courses/ all voc-ed courses 3. Voc-ed courses/100 students 4. Apprenticeship programs/ 100 students 5. Advanced voc-ed programs/100 students 6. Advanced courses/ advanced programs 7. Approved voc-ed programs/total programs SBCC .004 .628 10.608 .007 .288 23.116 .581 L.A. TradeTech .072 .193 12.150 .038 .218 10.765 .952 Mean .026 .265 4.796 .014 .188 6.128 .680 NOTE: Copper Mountain College is omitted. Minimum Maximum .000 .485 .008 .833 .927 22.060 .000 .101 .034 .788 1.750 38.444 .325 .970 17 comparable to that shown for credits in row 5 of Table 2.2. What we can do is express voc-ed course data on a per student basis, using CPEC Fall 2001 student headcounts. Row 3 of Table 2.3 indicates that SBCC and Los Angeles Trade-Tech each offer roughly 11 to 12 voc-ed courses per 100 students. Ratios of this magnitude look large (the mean is under five courses per 100 students), but the maximum ratio is over 22 courses per 100 students observed for Lake Tahoe Community College. Voc-Ed Program Measures Rows 4 and 5 of Table 2.3 make use of VTEA data measuring programs offered. Because VTEA data do not provide a reliable measure of total voc-ed programs, these rows report, respectively, apprenticeship and advanced occupational programs per 100 students. Row 4 continues the impression yielded by our credit and course data that apprenticeship programs are atypical. Row 5 offers a simple way to understand variation in advanced voc-ed programs, which is to “blow up” the ratios shown for a hypothetical college of 10,000 students. For such a hypothetical college, the extreme values suggest a sizable range of between 3 and 79 advanced programs offered. The minimum value of this ratio is observed for Palo Verde College and the maximum for Lassen College. Row 6 shows the “depth” or “intensity” of voc-ed programs measured as the ratio of advanced courses to advanced programs. For this measure, both Santa Barbara City College and Los Angeles TradeTech provide voc-ed programs of greater than average depth. Compared to the mean of about six advanced courses per advanced program, SBCC offers a ratio of 23 and Los Angeles Trade-Tech a ratio of nearly 11. The maximum value of over 34 advanced courses per advanced program appears for Cuesta College. In addition to VTEA data on voc-ed programs offered, the Chancellor’s Office maintains an “Inventory of Approved and Projected Programs.” The program inventory dataset includes all programs that are approved in the sense that a program must require 18 or more credits of coursework. In row 7 of Table 2.3, we use program inventory data to calculate for “approved” programs the ratio of voc-ed programs to all programs. This ratio for Los Angeles Trade-Tech is an astounding 95.2 18 percent. Even Santa Barbara City College reports that 58.1 percent of its total approved programs are voc-ed. Recall from Table 2.2 that the proportion of voc-ed credits to all credits for both colleges is about 25 percent. It is clear that voc-ed programs tend to be much smaller than other programs in terms of student credit hours. First-Time Freshman Data FTF data are derived from a longitudinal study following the universe of 222,372 first-time freshmen at all CCCS campuses between their initial enrollment in Fall 1997 and Spring 2000. We use FTF cohort data measuring students’ academic objectives formed without benefit of academic counseling to represent the demand of incoming students for community colleges’ academic programs. The data distinguish 14 categories of academic objectives, which we group into the following four broad categories of programs. Voc-ed 1. Formulate career interests 2. Vocational degree, no transfer 3. Acquire job skills 4. Update job skills 5. Vocational certificate 6. Maintain license Transfer 7. Degree and transfer 8. Transfer, no degree Basic skills 9. Basic skills 10. Education development 11. Complete GED Other 12. Degree, no transfer 13. Undecided 14. Uncollected 19 Data measuring academic objectives are available for 107 campuses. Again, Copper Mountain College is excluded. Of the four broad categories of programs distinguished for FTF data, probably the most clearly defined is interest in basic skills. As shown in row 1 of Table 2.4, the mean of the proportion of students with a basic skills objective is about 11 percent, which is reasonably close to the 7 percent shown in Table 2.2 for basic skills credits to all credits. Note that the denominator of this proportion excludes the “uncollected” category listed under “other” programs.1 No interest at all in basic skills is reported for freshmen at three colleges, including Canyons, Cerritos, and Citrus. Canyons College is highly unusual, however, because nearly 95 percent of freshman academic interests are undecided (see the Table 2.4 Measures of Curriculum Mix Using FTF Student Academic Objectives, 1997 Measure of Academic Objectives 1. Interest in basic skills SBCC .142 L.A. TradeTech .083 Mean Minimum Maximum .111 .000 .790 2. Interest in transferring .569 .111 .385 .033 .705 3. Interest in voc-ed .143 .644 .252 .017 .644 4. Interest in nontransfer programs .204 .686 .302 .018 .686 5. Undecided .086 .121 .202 .000 .949 NOTE: Copper Mountain College is omitted. _____________ 1Most campuses report either no data uncollected for their first-time freshmen or uncollected data for only a handful of students. However, FTF data are uncollected for more than 5 percent of first-time freshmen at 20 campuses; of these campuses, the uncollected rate is occasionally quite high. In particular, uncollected rates of 10 percent or higher are reported for 11 colleges, including at the high end Rio Hondo College (30.0%) and Sierra College (34.1%). We have no reason to expect that a systematic bias is causing these colleges to report high rates of uncollected responses. Therefore, we proceed with the proportions presented in Table 2.4 using as the denominator total students minus students for whom data could not be collected. 20 maximum in row 5). Minimum values shown in rows 2–4 are also observed for Canyon College. Taft College records the maximum value of the basic skills ratio (79%). We noted in discussing Table 2.2 that Taft College ranks very high in the proportion of voc-ed credits that are not transferable. An unexpected result in row 1 is that student interest in basic skills is slightly higher for Santa Barbara City College than for Los Angeles Trade-Tech. Row 2 of Table 2.4 reports the ratio of freshman students with an interest in transferring to a four-year institution, either with or without an A.A. degree. Excluding Canyon College, the minimum value for this ratio is 7.3 percent for Napa Valley College followed by Taft College at 8.5 percent. The maximum value of 70.5 percent is found for Citrus College. Transfer ratios calculated for Los Angeles Trade-Tech and Santa Barbara City College are close to the minimum and maximum values, respectively, indicating an even bigger difference between the two colleges in interest in transfer courses (45.8 percentage points) than in proportions of transferable credits actually taken (23.0 percentage points). FTF data allow only a narrow definition of voc-ed, which may well be appropriate because many freshman students are likely to be unaware that voc-ed courses are often transferable. In row 3, the mean voc-ed ratio is about one-quarter, with the maximum value of 64.4 percent appearing for Los Angeles Trade-Tech. Excluding Canyons College, the minimum value is for Taft College at 6 percent. This might seem surprising given what we know about Taft, but recall that nearly 80 percent of Taft freshman expect to enroll in basic skills programs. A huge 50 percentage point difference appears between the interest in voced of Los Angeles Trade-Tech freshman as opposed to Santa Barbara City College freshman, with only 14 percent of SBCC freshman expressing an interest in voc-ed programs. In row 4, we add students interested in an A.A. degree (but not in transferring) to those who express an interest in voc-ed. About 30 percent of all freshman report interest in this broader category of nontransfer programs, with the minimum value (excluding Canyons College) observed for Taft College and the maximum for Los Angeles Trade-Tech. 21 Finally, row 5 indicates that, on average, about 20 percent of freshmen students are undecided about their academic objectives. Undecided ratios calculated for both Santa Barbara City College and Los Angeles Trade-Tech are considerably below this mean. No students in the undecided category are observed for two campuses (Cerritos and Citrus), and the maximum value of 94.9 percent is, as noted, recorded for Canyons College. The next highest ratio is 59.9 percent calculated for San Francisco City College. Note that for the first three columns of Table 2.4, rows 1, 2, 4, and 5 sum to 100 percent, subject to rounding error. Summary This chapter describes our attempt at using CPEC campus descriptions to classify California community colleges by curriculum emphasis. This information allows us to measure curriculum mix by what colleges say about themselves in promotional materials. Our conclusion is that 28 out of 108 colleges indicated either a voc-ed or transfer emphasis. We next considered the wealth of quantitative data available online measuring curriculum in terms of credits, courses, programs, and student interests. In contrast to CPEC campus descriptions, these data allow measurement of what colleges actually offer their students. Descriptive statistics for curriculum mix variables constructed from the quantitative data indicate that, on average, nearly three-quarters of all credits in courses attempted are transferable. Measured at the extremes of the distribution, there appears to be considerable variation about this mean, as is the case for other curriculum measures including emphasis on voced, type and transferability of voc-ed offerings, and apprenticeship training. We also noted sizable differences in the mix of voc-ed offerings and interests of freshmen students between Los Angeles Trade-Tech and Santa Barbara City College—our “representative” colleges with voc-ed and transfer emphases, respectively. Chapter 3 examines whether these differences in curriculum mix are pervasive across colleges or are limited to colleges at the extremes. 22 3. Curriculum Specialization This chapter uses the quantitative data described in the previous chapter to identify colleges that differ in curriculum emphases as well as those that offer different curriculum specializations. We define differences in curriculum emphasis as systematic deviations in the curriculum offered by individual colleges from that of a typical or average college. Differences in curriculum specialization couple emphasis on one curriculum variable with a de-emphasis on another variable. For example, a college would specialize if it emphasizes a transferable curriculum while de-emphasizing nontransferable voc-ed. Note that all colleges that specialize must also have an emphasis, but not all colleges that have an emphasis meet our definition of specialization. In the first of four sections, we begin by describing differences between colleges in their curriculum offerings using histograms. Our purpose is to see whether the large spreads observed for many of our curriculum variables are due to only a small number of colleges lying far from means or whether a more substantial number of colleges deviate systematically from means. In the next section, we attempt to distinguish those curriculum variables for which colleges differ substantially in their emphasis. The tool we use to measure differences in emphasis is the interquartile (IQ) range, which contrasts the top 25 percent of colleges with the bottom 25 percent. In the third section, our definition of specialization is implemented by examining the relationships between selected curriculum measures making use of scatter diagrams and pair-wise correlations. This analysis is carried out for selected curriculum measures. The fourth section highlights results of a factor analysis of all of our curriculum measures intended to provide further evidence on curriculum emphases and specializations. 23 Getting Started with Histograms To get a feel for the importance of intercollege differences in curriculums, it is useful to examine visually the distributions of two of our PFE curriculum measures—transferable credits to all credits and narrow (or nontransferable) voc-ed credits to all voc-ed credits. Figure 3.1 displays a histogram for transferable credits. The figure shows that colleges with proportions of transferable credits of 0.5 or less are highly atypical. Instead, the mass of the distribution lies in the range between 0.6 and 0.9. As indicated in Table 2.2, this variable has a mean of 0.733, with a substantial range between a minimum value of 0.187 and a maximum of 0.922. Figure 3.2 shows the histogram for our measure of nontransferable voc-ed credits to all voc-ed credits. Compared to Figure 3.1, the “spread” of this distribution is much larger, so that more colleges are found near the extremes of the distribution. Figure 3.1 indicates that California community colleges are similar in their emphasis on the transfer mission, but keep in mind that for many colleges, transfer credits include a heavy dose of transferable voc-ed coursework. For only a small number of colleges do we observe substantial deviations from the mean. On the other hand, Figure 3.2 illustrates that substantial deviations from 40 30 Frequency 20 10 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 Ratio Figure 3.1—Ratio of Transferable Credits to All Credits 1.0 24 Frequency 20 15 10 5 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 Ratio Figure 3.2—Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits the mean are much more common, suggesting that colleges differ in their emphasis on nontransferable voc-ed. What Do IQ Ranges Tell Us About Differences in Emphasis? Table 3.1 presents several commonly used measures of variability for our 21 curriculum mix variables, where, as in Figures 3.1 and 3.2, we rely on measures of variability to capture differences in curriculum emphasis across colleges. In the first column, we carry over from Tables 2.2 to 2.4 the maximum/minimum range. Standard deviation is reported next. In the remaining three columns, we add the upper and lower quartiles and the associated interquartile range. As a measure of variability, IQ range has an advantage over the standard deviation of not being sensitive to extreme observations. Based primarily on the IQ range, our main conclusion is that colleges differ in curriculum emphasis primarily in terms of the transferability and advanced level of their voc-ed curriculums. On the other hand, we find that differences in emphasis across colleges are small for measures of apprenticeship training and basic skills and “residual” courses. 25 Table 3.1 Variability in Curriculum Mix Measures Using Data for Credits, Courses, Programs, and Student Objectives Curriculum Measure 1. Nontransferable voc-ed credits/all voc-ed credits 2. Advanced occupational credits/all voc-ed credits 3. Advanced occupational courses/all voc-ed courses 4. Approved voc-ed programs/total approved programs 5. Advanced courses/ advanced programs 6. Voc-ed courses/100 students 7. Advanced occupational programs/100 students 8. Interest in transferring 1. Broadly defined voc-ed credits/all credits 2. Transferable credits/all credits 3. Academic transfer credits/all credits 4. Interest in nontransfer programs 5. Interest in voc-ed 6. Undecided interest 7. Interest in basic skills Max/Min Standard First Range Deviation Quartile A. High Variability .920 .198 .207 .647 .142 .101 .825 .158 .153 .645 .140 .576 36.694 4.576 3.429 21.133 4.576 2.465 .754 .114 .114 .672 .127 .296 B. Moderate Variability .758 .105 .162 .735 .106 .693 .742 .111 .547 .668 .109 .229 .672 .104 .183 .949 .114 .149 .790 .094 .055 Fourth Quartile .481 .283 .348 .771 7.357 5.675 .231 .476 .260 .788 .667 .372 .310 .244 .137 IQ Range .273 .182 .195 .196 3.929 3.210 .116 .180 .098 .095 .120 .126 .095 .082 26 Table 3.1 (continued) Max/Min Standard First Fourth IQ Curriculum Measure Range Deviation Quartile Quartile Range C. Low Variability 1. Nontransferable voc-ed credits/all credits .756 .093 .036 .100 .064 2. Apprenticeship credits/ all voc-ed credits .727 .073 .000 .007 .007 3. Basic skills credits/all creditsa .207 .038 .081 .119 .039 4. Residual credits/all creditsa .311 .055 .076 .139 .063 5. Apprenticeship courses/ all voc-ed courses .485 .061 .000 .019 .019 6. Apprenticeship programs/100 studentsa .101 .021 .000 .021 .021 aIndicates variables classified on the basis of a small maximum/minimum range. To delve into the details of Table 3.1, we began by placing in the low variability category three variables with small maximum/minimum ranges: namely, basic skills credits to all credits, residual credits to all credits, and apprenticeship programs per capita. These variables appear in rows C.3, C.4, and C.6. The remaining 18 measures are candidates for the high or moderate variability categories. Fifteen of these remaining 18 variables are measured as proportions, two are measured in per capita terms, and one is program depth measured as courses per program. Beginning with variables measured as proportions, we showed in Figure 3.2 the histogram representing proportion of nontransferable voc-ed credits to all voc-ed credits. In the first row of Table 3.1, the IQ range for this variable (0.273) is seen to be the largest of the IQ ranges shown for all 15 proportionate variables. Its interpretation is that the top 25 percent of colleges (or 27 colleges) report a proportion of nontransferable voc-ed credits to all voc-ed credits that exceeds the proportion offered by the bottom 25 percent by at least 27.3 percentage points. A total of five variables measured as proportions are 27 classified as having high variability, which we define as IQ ranges between 0.180 and 0.273. These variables include • Nontransferable voc-ed credits to all voc-ed credits; • Advanced occupational credits to all voc-ed credits; • Advanced occupational courses to all voc-ed courses; • Approved voc-ed programs to all approved programs; and • Student interest in transferring to all student interests. Seven more variables measured as proportions are classified as having moderate variability as indicated by IQ ranges between 0.082 and 0.142. Note that these variables include the proportion of transferable credits to all credits, the variable depicted in Figure 3.1. The final three variables expressed as proportions have IQ ranges of 0.064 or below, and we place them in the low variability category. Two of these three variables involve apprenticeship training. It is more difficult to classify the remaining three variables involving courses and programs because they are not measured as proportions that must lie between 0.0 and 1.0. Nevertheless, we suggest that the ratio of advanced courses to advanced programs belongs in the high variability category (row A.5). The reason is that voc-ed programs provided by colleges in the fourth quartile (7.4 advanced courses per advanced program) have at least twice the depth as those offered by colleges in the first quartile (3.4 advanced courses per advanced program). We make a similar argument for the per capita measures of voc-ed courses and advanced voc-ed programs (rows A.6 and A.7, respectively). Row A.6 indicates that fourth quartile colleges offer students at least 5.7 voc-ed courses per 100 students, whereas first quartile colleges offer at most 2.5 voc-ed courses per 100 students. Row A.7 suggests that, blown up to the level of a hypothetical college of 10,000 students, fourth quartile colleges offer students at least 23 advanced voc-ed programs, whereas first quartile colleges supply at most 11 advanced voc-ed programs. For both variables, in other words, colleges in the top quartile offer at least twice the number of voc-ed courses or at least twice the number of advanced voc-ed programs as colleges in the bottom quartile. 28 Evidence of Specialization We turn now to the question of whether differences between colleges in curriculum mix represent specialization or an emphasis on one activity with a de-emphasis on another. To implement this definition, we look for evidence of negative relationships between pairs of curriculum variables as measured by correlation coefficients. Negative correlation coefficients are calculated in the two-variable case if, when one variable is above its mean, the other tends to fall below its mean. Table 3.2 displays estimated correlation coefficients calculated for selected pairs of curriculum mix measures. Our focus is primarily on four sets of correlation coefficients, which we label Correlation Results 1–4. Correlation Results 1 and 2 form the basis of our discussion of specialization in this section. Correlation Results 3 and 4, which are only briefly discussed, play a more prominent role in the factor analysis summarized in the next section. Correlation Result 1: Transferable Curriculum vs. Nontransferable Voc-Ed We find a large negative correlation (-0.837) between transferable credits to all credits and narrow voc-ed credits to all credits. That is, colleges above average in transferable credits tend to be below average in narrow voc-ed credits. And colleges above average in narrow voc-ed credits tend to be below average on their transferable credits ratio. This correlation coefficient is reported in row 2 and column (1) of Table 3.2.1 The negative relationship between the transfer credit ratio and the narrow voc-ed ratio is shown graphically in the scatter diagram appearing _____________ 1Recall from Chapter 2 that the transfer credit ratio and the narrow voc-ed ratio are developed from our most inclusive dataset, which we referred to as Goal 3 PFE data. The four mutually exclusive categories of Goal 3 PFE data are transfer, narrow voc-ed, basic skills, and residual credits. A negative relationship between any two of these variables is expected, since proportions representing mutually exclusive categories must sum to 1.0. Hence, being above average on one variable must be offset by being below average on at least one other variable. Nevertheless, a large negative coefficient is not guaranteed. In particular, the correlation between the transfer credit ratio and the basic skills ratio is only –0.09 and not statistically significant. Among the selected variables considered in Table 3.2, the only statistically significant negative correlation coefficient involving basic skills credits is that with freshmen interest in transfer programs. The estimated coefficient is –0.272. 29 30 Table 3.2 Estimated Correlation Coefficients Between Selected Curriculum Mix Measures Curriculum Measure 1. Transferable credits/all credits 2. Nontransferable voc-ed credits/all credits 3. Academic transfer credits/all credits 4. Nontransferable voc-ed credits/all voc-ed credits 5. Advanced occupational credits/all voc-ed credits 6. Advanced occupational courses/all voc-ed courses 7. Advanced courses/ advanced programs 8. Interest in transferring 9. Interest in voc-ed (1) 1.000 –.837 .781 –.766 .237 .383 .139 .355 –.185 (2) 1.000 –.668 .730 –.191 –.284 –.039 –.260 .076 (3) (4) 1.000 –.369 0.000 .162 –.186 .292 –.309 .069 –.068 .415 –.075 –.202 .195 NOTE: Estimates statistically significant at the 5 percent level are in bold. (5) (6) (7) 1.000 .802 1.000 .598 .635 1.000 .167 .270 .292 –.070 –.287 –.236 (8) 1.000 –.494 (9) 1.000 Deviations of transferable credits to all credits in Figure 3.3, where deviations from the mean of transferable credits are measured vertically and deviations from the mean of nontransferable voc-ed are measured horizontally. The negative relationship between these two ratios is captured in the figure by a concentration of data points in the upper left and lower right quadrants. The two outlying data points appearing in the lower right quadrant are for Palo Verde College and Taft College, with Taft College being farther from the origin.2 Recall from Chapter 2 that Taft College places a heavy emphasis on vocational education, particularly nontransferable voc-ed. It is a moderately small college with 8,033 students. Palo Verde College .4 .2 0 –.2 –.4 –.2 0 .2 .4 .6 Deviations of nontransferable voc-ed credits to all credits Figure 3.3—Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Credits _____________ 2Similar outlying data points in the upper left quadrant are not possible because of the large mean of transferable credits to all credits (0.733) and the small mean of narrow voc-ed credits to all credits (0.087). Extreme values of these variables reported in Table 2.2 imply that the maximum deviation from the mean for the transferable credits ratio is 0.189, and the maximum deviation from the mean for the narrow voc-ed credits ratio is –0.087. 31 is a very small college (2,903 students) in Blythe, in the southeastern California desert.3 Continuing with Figure 3.3, it appears that the bulk of the data points are in an imaginary box centered at the origin and extending right and left and up and down by about 5 percentage points. In an attempt to pin down those colleges that specialize as opposed to those that do not, we define colleges that specialize in offering a transferable curriculum as those for which deviations of the transferable credits ratio exceed 0.05 and deviations of the nontransferable voc-ed credits ratio are less than –0.05 (in other words, data points that lie above and to the left of the upper left corner of the box). Similarly, colleges that specialize in offering a nontransferable voc-ed curriculum are those for which deviations of the nontransferable voc-ed credits ratio exceed 0.05 and deviations of the transferable credits ratio are less than –0.05. The names of colleges that, according to these definitions, specialize in a transferable curriculum and in nontransferable voc-ed are reported, respectively, in Tables 3.3 and 3.4. Beginning with Table 3.3, we list in column (1) the names (in alphabetical order) of the 19 colleges that meet our definition of specializing in a transferable curriculum (termed “Definition 1”). For purposes of comparison, we repeat from Table 2.1 the names of 11 colleges that we classified in Chapter 2 as having a transfer emphasis based on our reading of CPEC campus descriptions. These 11 colleges are listed in column (3). There is some overlap between colleges listed in the two columns, with five colleges appearing in common. But it is worth noting that Santa Barbara City College and Santa Monica College are missing from column (1). Parallel to Table 3.3, Table 3.4 lists colleges that meet our definitions of nontransferable voc-ed specialization and emphasis. Column (1) of Table 3.4 reports that just 10 colleges specialize in nontransferable voc-ed, meaning that they emphasize nontransferable voc-ed credits while de-emphasizing transferable credits. As in Table 3.3, _____________ 3In comparison, Santa Barbara City College and Los Angeles Trade-Tech are midsized with 14,913 and 15,630 students, respectively. Several colleges have more than 30,000 students, with Riverside Community College the largest at 34,042 students. 32 Table 3.3 Colleges Emphasizing and Specializing in a Transferable Curriculum, Alternative Definitions of Specialization Transfer Specialization Emphasis Based on Definition 1 (1) Chabot Coastline Cuyamaca El Camino Grossmont Hartnell Las Positas Los Angeles Harbor Los Angeles Pierce Marin Orange Coast Palomar Reedley San Diego Mesa San Francisco City San Joaquin Delta Sierra Skyline Southwest Definition 2 (2) Chabot Coastline Consumnes River Cuyamaca Cypress Diablo Valley El Camino Grossmont Hartnell Las Positas Los Angeles Harbor Los Angeles Pierce Marin Napa Valley Orange Coast Palomar Reedley San Diego Mesa San Francisco City San Joaquin Delta San Mateo Santa Barbara Santa Monica Sierra Skyline Southwestern CPEC Descriptions (3) Chabot Diablo Valley Fullerton Las Positas Los Angeles Harbor Los Angeles Pierce Marin Oxnard Santa Barbara Santa Monica Ventura NOTES: Definition 1 transfer specialization is an emphasis on transfer credits coupled with a de-emphasis on nontransferable voc-ed credits measured relative to all credits. Definition 2 transfer specialization is an emphasis on transfer credits coupled with a de-emphasis on nontransferable voc-ed credits measured relative to all voc-ed credits. 33 Table 3.4 Colleges Emphasizing and Specializing in Nontransferable Voc-Ed, Alternative Definitions of Specialization Nontransferable Voc-Ed Specialization Emphasis Based on Definition 1 (1) Allan Hancock Los Angeles Trade-Tech Los Medanos Monterey Peninsula Palo Verde Rio Hondo Santa Ana Santiago Canyon Taft Victor Valley Definition 2 (2) Allan Hancock Contra Costa Evergreen Imperial Valley Laney Los Angeles City Los Angeles Trade-Tech Los Medanos Monterey Peninsula Moorpark Oxnard Palo Verde Porterville Rio Hondo Santa Ana Santiago Canyon Taft Ventura Victor Valley CPEC Descriptions (3) Coastline Contra Costa Copper Mountain Consumnes River Desert East Los Angeles Foothill Fresno City Lassen Long Beach City Los Angeles Trade-Tech Moorpark San Diego Miramar San Joaquin Delta Shasta Siskiyous Taft NOTES: Definition 1 nontransferable voc-ed specialization is an emphasis on nontransferable voc-ed credits measured relative to all credits coupled with a de-emphasis on transfer credits. Definition 2 nontransferable voc-ed specialization is an emphasis on nontransferable voc-ed credits measured relative to all voc-ed credits coupled with a deemphasis on transfer credits. there is limited overlap between these 10 colleges and the 17 colleges listed in column (3) as having a voc-ed emphasis based on CPEC descriptions. Los Angeles Trade-Tech and Taft College appear in both columns, but Palo Verde College is listed in column (1) but not in column (3). 34 Correlation Result 2: Transferable Curriculum vs. Proportion of Voc-Ed That Is Nontransferable Making use of the detail available on the mix of voc-ed credits, we obtain a negative correlation coefficient (–0.766) between transferable credits to all credits and narrow voc-ed credits to all voc-ed credits.4 This negative relationship is reported in row 4 and column (1) of Table 3.2 and shown graphically in Figure 3.4. As in Figure 3.3, note in Figure 3.4 the concentration of data points in the upper left and lower right quadrants. A difference between the two figures is that the more narrowly defined denominator in the ratio of nontransferable voc-ed credits to all voc-ed credits has the effect of “spreading” horizontally the scatter diagram in Figure 3.4. Employing the imaginary box applied in Figure 3.3 with dimensions of 5 percentage points in each direction from the origin, we apply the same definitions of what it means to specialize in, say, a transferable curriculum by being above the mean of the transferable credits ratio by at least 5 percentage points while being below the nontransferable credits ratio by at least 5 percentage points. The greater horizontal spread of Figure 3.4 results in a larger number of colleges that meet this definition of specializing (termed “Definition 2”). Names of the 26 colleges that specialize in a transferable curriculum according to Definition 2 appear in column (2) of Table 3.3. Note that both SBCC and Santa Monica College are listed. Column (2) of Table 3.4 lists the 19 colleges that specialize in a nontransferable voc-ed curriculum according to the information included in Figure 3.4. Our preferred definition of specialization is the more inclusive definition implemented in column (2) of both tables. The reason is that the column (2) definition incorporates the breadth of Goal 3 PFE data covering transferable, voc-ed, and basic skills curriculums with the detail on voc-ed offerings available in Goal 4 PFE data. _____________ 4As described in Chapter 2, PFE Goal 4 data on the mix of voc-ed credits allow us to measure nontransferable voc-ed credits as a proportion of all voc-ed credits. 35 Deviations of transferable credits to all credits .4 .2 0 –.2 –.4 –.4 –.2 0 .2 .4 Deviations of nontransferable voc-ed credits to all credits .6 Figure 3.4—Ratio of Transferable Credits to All Credits vs. Ratio of Nontransferable Voc-Ed Credits to All Voc-Ed Credits Correlation Result 3: Positive Relationships Between Measures of Advanced Voc-Ed Curriculum Adding VTEA data for courses and programs to our PFE measures of credits, we obtain a cluster of three positive correlation coefficients relating variables measuring mix of advanced voc-ed credits, mix of advanced voc-ed courses, and advanced voc-ed courses to programs. The three positive correlation coefficients, which appear in rows 6 and 7 and columns (5) and (6) of Table 3.2, range in value from 0.598 to 0.802. Correlation Result 4: Tradeoff Between Student Interests in Voc-Ed and Transferring Our final correlation result introduces FTF measures of freshmen student academic interests. We find a sizable negative relationship between student interest in voc-ed and student interest in transferring. As seen in row 9 and column (8) of Table 3.2, the correlation coefficient between these two variables is –0.494. It is worth noting from row 9 that student interest in voc-ed is also negatively correlated with measures 36 of academic transfer credits (-0.202), advanced voc-ed courses (-0.287), and advanced courses to advanced programs (-0.236). Bringing It Together Using Factor Analysis The findings in the previous section are based on an analysis of measured relationships between nine of our curriculum variables. In this section, we buttress these findings with results obtained using a more rigorous, formal approach called factor analysis, which we apply to all 21 of our curriculum variables.5 Briefly, the objective of factor analysis is to help researchers categorize their data. More specifically, factor analysis identifies a small number of underlying unobserved variables, called factors, that account for measured relationships between a larger number of observed variables. Our results indicate that the relationships between our curriculum mix variables can be captured by just four unobserved factors. Ranked in descending order by their ability to distinguish colleges in terms of observed curriculum measures, the four factors and the labels we assign to each are as follows: Factor 1: transferable curriculum/nontransferable voc-ed tradeoff; Factor 2: advanced voc-ed curriculum; Factor 3: demand for nontransferable curriculum; and Factor 4: apprenticeship training. Allowing us to identify these four factors is an underlying matrix of factor loadings that we estimate to measure the importance of each observed variable in explaining a particular factor. (A detailed description of factor analysis and our key results is presented in Appendix B.) Beginning with Factor 1, the message we take from our results is a distinction between colleges that emphasize a transferable curriculum _____________ 5We actually used 18 of our 21 curriculum variables in the factor analysis. The three omitted variables are residual credits to all credits, first-time freshmen undecided about their academic objectives to all freshmen, and broadly defined voc-ed credits to all credits. These variables are omitted because they are linear combinations of the remaining variables. 37 and colleges that emphasize nontransferable voc-ed and appeal to students interested in basic skills. Large positive factor loads are obtained for measures of transferable credits to all credits and academic transfer credits to all credits. Negative factor loads appear for nontransferable voc-ed credits to all credits, nontransferable voc-ed credits to all voc-ed credits, and student interest in basic skills. In other words, Factor 1 appears to capture the two curriculum specializations we described in the previous section based on our Correlation Results 1 and 2. The first of these is a transfer curriculum specialization, and the second is specialization in nontransferable voc-ed. We label Factor 2 advanced voc-ed curriculum, as our data seem to be contrasting colleges that emphasize an advanced vocational curriculum with all other colleges. Large and positive factor loadings are observed for three variables: advanced voc-ed credits to all voc-ed credits, advanced occupational courses to all voc-ed courses, and advanced occupational courses to advanced occupational programs. Recall that positive correlation coefficients between these three variables were noted above as Correlation Result 3. Positive factor loads for Factor 3 are obtained for FTF measures of student interest in voc-ed, student interest in more broadly defined nontransferable programs (i.e., interest in voc-ed plus interest in an A.A. but not in transferring), and proportion of credits in basic skills. At the same time, a negative factor load appears for student interest in transfer programs. Since the variables with large factor loads are primarily calculated from FTF measures of student objectives, we label Factor 3 demand for nontransferable curriculum. Consistent with Correlation Result 4, these results suggest a distinction between colleges whose students express an interest in nontransferable programs and colleges whose students are interested in other curriculums, both voc-ed and traditionally academic, that are transferable to a four-year college. Factor 4 is clearly apprenticeship training. Large and positive factor loadings are observed for our measures of apprenticeship courses to all voc-ed courses, apprenticeship credits to all voc-ed credits, and apprenticeship programs per 100 students. Estimated factor loadings can be used to calculate predicted values of the common factors, called factor scores, for each community college in 38 our dataset. Thus, for each factor, factor scores allow us to measure the magnitude of differences between individual colleges. Table 3.5 lists the top five and bottom five colleges ordered by factor scores for each of the four factors. Beginning with Factor 1, the transferable curriculum/ nontransferable voc-ed tradeoff, both top-ranked Orange Coast College and bottom-ranked Taft College were mentioned in Chapter 2 as being at the extremes of the distribution for the PFE measure of transferable credits. All five top-ranked colleges appear in the Table 3.3 list of colleges that specialize in a transfer curriculum, using either the column (1) or column (2) definitions. Only Lassen College among the five bottom-ranked colleges does not appear in the Table 3.4 list of colleges that specialize in nontransferable voc-ed. With respect to the remaining factors, it is interesting to note for Factor 2 that Santa Barbara City College and Santa Monica College rank in the top five in terms of offering advanced voc-ed, although they do not rank in the top five for specializing in a transferable curriculum. The Table 3.5 Top Five and Bottom Five Colleges Ordered by Factor Scores for Factors 1–4 Factor 1: Transferable Curriculum/ Nontransferable Voc-Ed Tradeoff Orange Coast Palomar San Diego Mesa Sierra Los Angeles Pierce Taft Palo Verde Lassen Los Medanos Rio Hondo Factor 3: Factor 2: Demand for Factor 4: Advanced Voc-Ed Nontransferable Apprenticeship Curriculum Curriculum Training Top Five Colleges Cuesta Los Angeles Santiago Canyon Santa Barbara Southwest San Mateo Galivan Los Angeles Trade- Los Angeles Trade- Santa Monica Tech Tech Rio Hondo Los Angeles Mission Rio Hondo San Jose City Foothill Compton Bottom Five Colleges Santiago Canyon Taft Taft Marin Diablo Valley Siskiyous Merced Orange Coast Lassen West Hills DeAnza Canyon Palo Verde West Valley Merritt 39 top-ranked college for Factor 2, Cuesta College, was noted in Chapter 2 as being at the extreme right-hand tail of the distribution of advanced courses per advanced program. Summary Do California community colleges specialize in the academic programs they offer? Our answer, using three different empirical approaches, is that many do. We first examined interquartile ranges for our 21 curriculum mix variables to check whether large differences identified at the extremes of the distributions hold up more broadly across colleges. Our results indicated sizable differences between the top 25 percent and the bottom 25 percent of colleges for eight variables that are primarily measures of advanced level or transferability of voc-ed curriculums. Next we examined relationships between pairs of curriculum variables to determine whether intercollege differences can be interpreted as specialization. Using scatter diagrams to visualize pair-wise correlation coefficients between variables, we implemented a common-sense definition of specialization linking emphasis on one curriculum measure with a de-emphasis on another. Our preferred specification indicates that 26 colleges offer a transfer specialization, and 19 colleges specialize in nontransferable voc-ed. The correlation coefficients also suggested (1) a direct relationship between various measures of advanced voc-ed, and (2) a negative relationship between student interest in voc-ed and in other types of programs. Our third empirical approach used factor analysis to identify unobserved common factors that underlie observed relationships among all of our curriculum measures. We found that our data can be summarized by four common factors. Of the four factors, Factors 1-3 identify differences between colleges that reinforce three important results coming out of our pair-wise correlations. These are that (1) some colleges specialize in offering a transferable curriculum and others specialize in nontransferable voc-ed, (2) colleges differ in their emphasis on advanced voc-ed, and (3) colleges differ in the interest of their students in nontransferable voc-ed programs. 40 4. Explaining Differences in Curriculum Mix Chapter 3 presented evidence suggesting that California community colleges differ in terms of their curriculum emphases and specializations. In this chapter, we ask whether these differences can be explained. To answer this question, we append to our dataset measuring college curriculum variables additional data measuring college-specific characteristics, local job opportunities, and community demographics. The first section provides a brief overview of our measures of collegespecific characteristics, local job opportunities, and demographic characteristics. Appendix C includes a description of the online data sources and technical details involved in constructing these variables. A complete listing of all variables included in our dataset appears in Appendix D. Our main substantive results are reported in the second section, where we present cross-tabulation evidence indicating that differences between colleges in curriculum emphasis and specialization can be reasonably explained by our measures of college-specific characteristics and community needs. Description of Explanatory Variables College-Specific Characteristics Measures of campus-specific characteristics are broken up into two categories that we term institutional variables and student demographics. For the first category, Table 4.1 presents simple descriptive statistics for measures of membership in a multicampus district, proximity to a state four-year college, and campus age. Of the 108 CCCS campuses, 56 campuses (51.9 percent) are in 21 multicampus districts. The number of campuses included in each of these multicampus districts is not reported 41 Table 4.1 Descriptive Statistics for Explanatory Variables Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Age of campus (years) Student demographics Asian Black Filipino Hispanic Native American White Nonresident alien Other Nonresponse Female Male Gender unknown Employer characteristics Percentage of employees in Manufacturing Wholesale trade Retail trade Real estate Professional services Administration and support services Education services Health care Arts and entertainment Accommodations and food services Other services (except public administration) Total employees (thousands) Community demographics Bachelor’s degree Foreign born Standard Mean Deviation Minimum Maximum .519 — .000 1.000 47.5 51.4 26.9 30.9 53.1 22.6 1.7 282.5 1.5 178.2 1 119 .098 .090 .077 .097 .031 .028 .253 .156 .011 .009 .438 .199 .013 .014 .017 .015 .062 .047 .558 .062 .435 .064 .006 .010 .010 .003 .002 .039 .002 .021 .000 .000 .000 .200 .318 .000 .424 .694 .159 .856 .064 .869 .091 .178 .237 .677 .798 .054 .165 .123 .083 .057 .216 .087 .030 .020 .074 .048 .108 .063 .005 .004 .106 .049 .023 .029 .151 .064 .000 .002 .030 .008 .011 .013 .000 .009 .000 .030 .533 .400 .436 .179 .218 .322 .030 .256 .179 .398 .039 .014 76.3 178.4 .265 .146 .233 .116 .000 .085 .3 898.9 .054 .781 .013 .544 42 Variable Median household income ($ thousands) Race White Black Native American Asian Other race Two or more races Hispanic ethnicity Table 4.1 (continued) Standard Mean Deviation Minimum Maximum 49.3 21.2 21.9 173.6 .621 .167 .063 .075 .014 .008 .112 .110 .142 .104 .048 .013 .281 .174 .167 .002 .002 .003 .005 .019 .022 .937 .403 .045 .618 .494 .101 .893 in the table. The Los Angeles district is by far the largest district with nine campuses. The Peralta district serving the Oakland metropolitan area is second largest with four campuses. The next two rows of Table 4.1 measure proximity to the nearest UC and CSU campuses. The table indicates large variation about the means of 47.5 miles to the nearest UC campus and 26.9 miles to the nearest CSU campus. At the extreme of easy access, it is only 1.7 miles from Irvine Valley College to the nearest UC campus, and only 1.5 miles from San Francisco City College to the nearest CSU campus. At the other extreme, College of the Redwoods on the Northern California coast in Eureka is nearly 283 miles from the nearest UC campus, and Palo Verde College in the southeastern California desert, is 178 miles from the nearest CSU campus. Recall that Palo Verde College was described in Chapter 3 as a very small college that places heavy emphasis on nontransferable voc-ed. Campus age measured from 2002 is widely dispersed, but there is some concentration in the data at founding dates in the early to mid1950s. Chaffey College, founded in 1883, is the oldest CCCS institution; Copper Mountain College, founded in 2001, is the youngest. The student demographic variables in Table 4.1 include race and ethnicity as well as gender. Kane and Rouse (1999) report that at the national level, the combined student body of community colleges is 70 percent white, 11 percent black, and 11 percent Hispanic. (Kane and 43 Rouse do not break out the percentage of Asian students.) It is apparent that California community colleges enroll on average a much higher percentage of Hispanic students (25.3 percent) and a much lower percentage of white students (43.8 percent) than do community colleges nationally. Indeed, nine campuses mostly in the Los Angeles Basin and south to San Diego fall into the category of “predominantly Hispanicserving institutions” (PHIs), defined as those whose student bodies are over 50 percent Hispanic. A total of 47 community colleges, including the PHIs, fall into the more numerous “Hispanic-serving institutions” (HSIs) category, defined as those whose student bodies are 25 percent or more Hispanic. Imperial Valley College near the Mexican border reports the maximum proportion of Hispanic students (85.6 percent). Regarding gender, the unexpectedly low minimum ratio of female students (just 20 percent) is observed for Taft College. Local Job Opportunities Shown next in Table 4.1 are descriptive statistics for our two measures of local job opportunities—industry mix and total employment. Grubb (1996) emphasizes that the labor market for community college students (what he calls the “sub-baccalaureate labor market”) is almost entirely local. To concentrate on local labor markets, we make use of 1997 Economic Census information for employers in the immediate proximity of the community college campus using a threestep protocol described in Appendix C. With this protocol, we obtained unique college-specific employment information for 94 colleges. The remaining 14 colleges are all in large metropolitan areas, including Los Angeles, Oakland, Sacramento, San Diego, and San Jose. We measure employer characteristics using data for the entire city for at least two colleges in each of these metropolitan areas. In terms of industry mix, Table 4.1 shows that the largest spread between maximum and minimum values appears for manufacturing, followed by retail trade, wholesale trade, and accommodations and food services. Also observed is a huge variation in the size of the local labor market as measured by total employment. Total employment ranges between 335 employees for Foothill College in Los Altos Hills and nearly 44 900,000 employees for the four Los Angeles area colleges for which we use data for the Los Angeles metropolitan area. Community Demographics The community demographic variables appearing at the bottom of Table 4.1 are based on 2000 Census data for the city in which the college is located using the protocol developed for measuring local employer characteristics. The variation shown in proportions of bachelor’s degree holders and foreign-born residents and in median household income illustrates the diversity of California’s population. Maximum and minimum values for bachelor’s degree are obtained for Foothill College (78.1%) and Los Angeles Mission College (5.4%), respectively. The maximum value of median household income is also reported for Foothill College (over $173,000). The proportion of foreign-born residents ranges between the minimum reported for Lassen College (1.3%) and the maximum observed for Glendale Community College (54.4%). Lassen College is in the Sierra Nevada region of northeastern California; Glendale Community College is in the greater Los Angeles metropolitan area. Our remaining community demographic variables include race and ethnicity, each measured separately. The race/ethnicity mix of students may differ somewhat from the race and ethnicity of the community because colleges draw students from outside the geographic boundaries of their communities. Nevertheless, there appears to be a direct correspondence between the two. As one example, Feather River College reports the maximum proportion of white residents in its local community (93.7%) and the maximum proportion of white students (86.9%). Maximum values for other major race and ethnicity variables measured at the community level are 40.3 percent for blacks at Compton College, 61.8 percent for Asians at East Los Angeles College, and 89.3 percent for Hispanics at Los Angeles Mission College. The correlation between the racial and ethnic composition of students and community race and ethnicity is strong: We estimate correlation coefficients between corresponding student and community variables of 0.76 for blacks, 0.69 for Asians, 0.76 for Hispanics, and 0.76 for whites. 45 Explaining Differences in Specializations and Emphases We turn now to the question of the extent to which these explanatory variables may be helpful in understanding intercollege differences in curriculum specializations and emphases. Specializations in Transferable Curriculums and Nontransferable Voc-Ed Our two curriculum specialization measures are developed from a scatter diagram relating the ratio of transferable credits to all credits to the ratio of voc-ed credits that are nontransferable (see Figure 3.4). Colleges that specialize in a transferable curriculum are those above the mean of the transferable credits ratio by at least 5 percentage points but below the mean of the nontransferable voc-ed credits ratio by at least 5 percentage points. Similarly, colleges that specialize in nontransferable voc-ed are those above the mean of the nontransferable voc-ed credits ratio by at least 5 percentage points but below the mean of the transferable credits ratio by at least 5 percentage points. Column (1) of Table 4.2 displays means of selected explanatory variables for the 26 colleges listed in Table 3.3 as having a transfer curriculum specialty using Definition 2. Column (2) provides the same information for the 19 colleges listed in Table 3.4 that specialize in nontransferable voc-ed. Means of explanatory variables for the remaining 63 colleges appear in column (3). The first group of explanatory variables shown in Table 4.2 consists of campus-specific institutional variables. We expect to see colleges in multicampus districts offering a more specialized curriculum than colleges in single-campus districts. A college in a multicollege district must compete with other colleges to enroll local students, and one way to compete is for the college to differentiate itself in terms of its mix of academic programs. On the other hand, colleges that are the sole provider within reasonable commuting distance do not face this kind of competition and may be expected to offer a broad range of educational services to satisfy the diverse needs of local residents and employers. 46 Table 4.2 Cross-Tabulations Between Specializations in Transferable Curriculums and Nontransferable Voc-Ed and Selected Explanatory Variables Explanatory Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Student demographics Asian Black Hispanic White Employer characteristics Manufacturing employment Total employees (thousands) Community demographics Bachelor’s degree Foreign born Median household income ($ thousands) Race White Black Asian Hispanic ethnicity Transfer Specialization (1) Nontransferable Voc-Ed Specialization (2) Other Colleges (3) .577 .632 .460 25.0 54.7 54.6 15.6 33.9 29.4 .120 .082 .094 .058 .081 .084 .219 .343 .239 .451 .359 .457 .159 86.1 .185 128.9 .152 56.5 .299 .204 .269 .250 .264 .217 51.6 46.1 49.3 .649 .564 .627 .043 .074 .068 .116 .081 .120 .265 .411 .248 Number of colleges 26 19 63 Table 4.2 demonstrates that, consistent with these expectations, colleges that specialize are more likely than other colleges to be part of multicampus districts. In particular, columns (1) and (3) indicate that colleges with a transfer specialization are nearly 12 percentage points more likely to belong to a multi-campus district than are “other” colleges. Similarly, columns (2) and (3) show that colleges with a nontransferable voc-ed specialization are 17 percentage points more likely to belong to a multicampus district than are other colleges. 47 Consider next proximity to a four-year college. For at least two reasons, colleges close to a four-year college are more likely than other colleges to specialize in a transferable curriculum. First, proximity makes transferring cheaper for students because they may be able to continue to live with their parents or in their own homes. Second, proximity is likely to lead to closer relationships between faculties of the community college and the four-year college, which, in turn, should result in a greater number of articulation agreements and a more seamless transfer process. Again, results shown in Table 4.2 are consistent with expectations. A comparison of columns (1) and (2) shows that colleges that specialize in a transferable curriculum are on average nearly 30 miles closer to a UC campus and about 18 miles closer to a CSU campus than are colleges with a nontransferable voc-ed specialization. Similarly, columns (1) and (3) show that colleges with a transferable curriculum specialization are over 29 miles closer to a UC and about 14 miles closer to a CSU than are other colleges. In other words, colleges offering a transferable curriculum are on average about half the distance to the nearest UC or CSU campus compared to colleges with a nontransferable voc-ed specialization and all other colleges. Within the other college category, we singled out for further analysis 17 colleges offering what might be termed a “comprehensive” curriculum, by which we mean a mix of transferable credits and nontransferable voc-ed credits that is within 5 percentage points of the mean of each variable. Probability of membership in a multicampus district is 41 percent for these colleges compared to the 46 percent shown in Table 4.2 for all 63 other colleges. Consequently, gaps in multicampus membership between colleges with a curriculum specialization and comprehensive colleges are even larger than the gaps appearing in the table. For example, colleges with a transfer specialization are nearly 17 percentage points more likely than comprehensive colleges to belong to a multicampus district, compared to the 12 percentage point gap in Table 4.2 between colleges with a transfer specialization and other colleges. Similarly, comprehensive colleges are even farther away from the nearest UC and CSU campuses than are the 63 other colleges—about 65 miles to the nearest UC and 39 miles to the nearest CSU. 48 Regarding race and ethnicity of students, Table 4.2 suggests that Asian students are slightly more likely to attend colleges with a transfer specialization than colleges specializing in nontransfer voc-ed. On average, Asian students represent 12 percent of students attending colleges with a transfer specialization, whereas 8 percent of students attending colleges with a nontransfer voc-ed specialization are Asian. Among the other race and ethnicity categories, columns (1) and (2) indicate a larger difference in the opposite direction for Hispanic students. Hispanics are over 12 percentage points more likely to attend a college specializing in nontransferable voc-ed than to attend a college with a transfer specialty. Turning to employer characteristics, Table 4.2 indicates that colleges specializing in nontransferable voc-ed tend to be in larger labor markets than are colleges with a transfer specialty. In addition, both categories of colleges that specialize are in larger labor markets than are the remaining 63 colleges. Differences in mix of employment, in contrast, tend to be quite small. The table shows that the percentage of local employment in manufacturing is higher by less than 3 percentage points for colleges that have a nontransferable voc-ed specialty than it is for colleges with a transfer specialty. Finally, the community demographic variables shown in Table 4.2 suggest that colleges with a transfer specialization are somewhat more likely to be in more highly educated, higher-income communities than are colleges with a nontransferable voc-ed specialization. For example, columns (1) and (2) indicate a nearly 10 percentage point difference in the percentage of residents with a bachelor’s degree. Echoing an earlier result for Hispanic students is the finding that communities with colleges that specialize in nontransferable voc-ed have a 15 percentage point greater share of Hispanics than do communities with colleges with a transfer specialization. Emphasis on Advanced Voc-Ed In Chapter 3, we noted as Correlation Result 3 the existence of large positive relationships between three measures of advanced voc-ed: (1) advanced voc-ed credits to all voc-ed credits, (2) advanced voc-ed courses to all voc-ed courses, and (3) ratio of advanced voc-ed courses to 49 advanced voc-ed programs. Any one of these variables would serve as a reasonable candidate for measuring an advanced voc-ed emphasis. The particular advanced voc-ed variable we examine in Table 4.3 is advanced voc-ed courses to all voc-ed courses. The first two columns of the table contrast colleges in the top and bottom quartiles of the ratio of advanced voc-ed courses to all voc-ed courses for the same selected explanatory variables appearing in Table 4.2. Table 4.3 Cross-Tabulations Between Emphasis on Advanced Voc-Ed Courses and Student Interest in Voc-Ed and Selected Explanatory Variables Explanatory Variable Institutional variables Multicampus district Proximity to four-year college Miles to nearest UC Miles to nearest CSU Student demographics Asian Black Hispanic White Employer characteristics Manufacturing employment Total employees (thousands) Community demographics Bachelor’s degree Foreign born Median household income ($ thousands) Race White Black Asian Hispanic ethnicity Number of colleges Emphasis on Advanced Voc-Ed Top Quartile (1) Bottom Quartile (2) Student Interest in Voc-Ed Top Quartile (3) Bottom Quartile (4) .462 .333 .630 .667 31.9 73.5 35.7 39.5 17.6 41.6 27.4 19.6 .092 .070 .094 .123 .052 .078 .120 .054 .273 .227 .310 .200 .446 .513 .333 .482 .183 61.4 .124 93.3 .089 167.3 .092 50.8 .270 .210 .251 .333 .229 .199 .282 .232 49.9 40.3 52.1 54.0 .641 .630 .533 .680 .045 .080 .082 .037 .103 .077 .132 .126 .307 .276 .348 .211 26 27 27 27 50 Comparing columns (1) and (2) of Table 4.3, the primary differences between colleges in the top and bottom quartiles are for the campusspecific institutional variables. In particular, colleges emphasizing advanced voc-ed are 13 percentage points more likely than colleges that do not emphasize advanced voc-ed to be part of a multicampus community college district. In addition, top-quartile colleges for this variable are nearly 42 miles closer to a UC campus and 24 miles closer to a CSU campus than are bottom-quartile campuses. If we pull together comparable estimates in Tables 4.2 and 4.3, the picture that emerges is that colleges with a transfer specialization or an emphasis on advanced voc-ed are more likely to be part of a multicampus district and to be within reasonable commuting distance to a UC or CSU campus. Differences between colleges in emphasis on advanced voc-ed tend to be small for race and ethnicity, whether measured for students or for the community. However, the limited evidence in Table 4.3 suggests that emphasis on advanced voc-ed is somewhat more common for colleges in more highly educated, higher-income communities. For example, the incidence of bachelor’s degrees is 6 percentage points higher for top quartile than for bottom quartile colleges. Top quartile colleges also tend to be in somewhat smaller labor markets and in labor markets with a higher proportion of manufacturing employment. Student Interest in Voc-Ed The third difference between colleges singled out for further attention in Chapter 3 is in freshmen student interest in voc-ed programs. We also mentioned in Chapter 2 that student interest in voced is likely to be primarily capturing demand for nontransferable voc-ed. Columns (3) and (4) in Table 4.3 compare top and bottom quartiles of colleges measured in terms of student interest in voc-ed. It should be noted that colleges appearing in the bottom quartile, indicating a low level of student interest in voc-ed, are likely to be colleges whose students exhibit a high level of interest in transfer programs. A comparison of columns (3) and (4) in Table 4.3 indicates small differences between top and bottom quartile colleges for membership in a multicampus district and proximity to the nearest four-year college. 51 Larger differences between quartiles appear for student demographic variables. In particular, proportions of black and Hispanic students are 7 percentage points and 11 percentage points higher, respectively, for top quartile than for bottom quartile colleges. On the other hand, colleges with large proportions of Asian and especially white students are more likely to enroll students with relatively little interest in voc-ed. Race and ethnicity results at the community level echo these findings. For instance, the proportion of Hispanic residents is nearly 14 percentage points higher for top quartile than for bottom quartile colleges. Student interest in voc-ed is also positively related to proportion of foreign-born residents in the community and to the size of the labor market and negatively related to the proportion of bachelor’s degree holders (Table 4.3). Summary Colleges that specialize in either a transferable curriculum or in nontransferable voc-ed are more likely than other colleges to be part of a multicampus district and to be in large labor markets. Between these two categories of specialization, colleges that have a transfer specialization are in closer proximity to a UC or CSU campus, whereas colleges that specialize in nontransferable voc-ed tend to have a larger proportion of Hispanic students and to be in larger communities with greater minority populations. We also compared colleges that differ in their emphasis on advanced voc-ed curriculums and in the interest of their freshmen students in voced programs. Emphasis on advanced voc-ed is directly related to membership in a multicampus district, proximity to a UC or CSU campus, and proportion of manufacturing employment. The racial and ethnic mixes of the student body and the community have a stronger effect on student interest in voc-ed programs, with interest in voc-ed increasing with the proportions of Hispanics attending the college or residing in the community. Demand for voc-ed programs also increases with size of the local labor market. 52 5. Policy Discussion The purpose of this concluding chapter is to provide guidance to policymakers for evaluating the performance of community colleges. The chapter is divided into three sections. The first reviews key provisions of the California Master Plan for Higher Education. The second provides some detail on the multiple performance indicators faced by community colleges. The third discusses several observations drawn from our empirical findings that we suggest should be taken into account when evaluating California community colleges. Key Provisions of the California Master Plan The 1960 California Master Plan laid the foundation for the subsequent development of the California Community College System on two guiding principles. The first is a policy of open admission to state high school graduates at no tuition (later, that principle was modified to very low tuition). California still leads the nation in accessibility of community college educational services and low tuition (Romano, 2003). The second guiding principle is the flexibility granted to community college districts and individual colleges to define their own missions. As stated in the Master Plan, community colleges are to offer instruction in (1) standard collegiate courses for transfer to fouryear institutions, (2) vocational-technical fields leading to employment, and (3) general or liberal arts courses (California State Department of Education, 1960, p. 36). Hence, community college districts and individual colleges have for decades enjoyed considerable latitude in setting their missions in line with perceived local business and community needs and with the vision of college leaders. Over time, the Master Plan has frequently been amended by the California state legislature. An important amendment was passed in 1996, reiterating that the primary missions of California community colleges are to offer traditional academic programs and vocational 53 instruction at the lower-division level. Academic and voc-ed programs are to be accessible to both traditional college-age and older students, including adults returning to school. As part of these primary missions, the amendment specifies that colleges are to offer remedial instruction and adult noncredit education. The 1996 amendment states that an additional primary mission of community colleges is “to advance California’s economic growth and global competitiveness through education, training, and services that contribute to continuous workforce improvement” (State of California, 1996, section 2). To help understand the legislature’s intent, other provisions of the amendment add that in fulfilling this new mission, community colleges are to play a more active and even entrepreneurial role in local and regional economic development. Along with this emphasis on workforce development, the legislature also established in 1996 the California Performance Based Accountability (PBA) system to monitor the performance of community colleges in meeting workforce development goals. Performance Evaluation Requirements As their missions have expanded, California community colleges have also been faced with increasing demands for accountability in their use of taxpayers’ money. Chapter 1 described the federal student-rightto-know regulation requiring that colleges collect and report student transfer and completion rates. As we noted above, data for First-Time Freshman student cohorts are used to calculate SRTK completion and transfer rates. More important are the reporting requirements established by two additional pieces of federal legislation, both of which were passed in 1998. The Workforce Investment Act specifies a total of 17 core indicators of performance and consumer satisfaction for state institutions, including community colleges. Core indicators of performance include such labor market outcomes as placement in unsubsidized employment, initial wages, and retention in employment and wages measured after six months on the job. These employmentrelated measures are to be obtained from Unemployment Insurance (UI) quarterly earnings histories. Levels of performance on core indicators 54 that community colleges are expected to meet or exceed are established in negotiations between federal and state officials. Specific to voc-ed programs, VTEA establishes a similar set of core indicators of performance and requires that states collect and report data with which performance can be assessed. Not covered under WIA is an additional VTEA goal requiring completion of voc-ed programs that lead to nontraditional training and employment. Performance goals required by WIA were established in the 1996 Partnership for Excellence agreement between the State of California and the CCCS. The five PFE goals include • A greater number of transfers to UC and CSU campuses; • An increased number of degrees and certificates awarded; • Higher rates of course completion for transfer, voc-ed, and basic skills courses; • Greater contribution to workforce development as measured by completion of apprenticeship, advanced voc-ed, and introductory voc-ed courses; and • Basic skills improvement as measured by number of students completing coursework at least one level above prior basic skills courses. The core indicators required by federal VTEA legislation are specified in the 1994 California Perkins State Plan. These core indicators include • Completion rates for courses in voc-ed programs; • Percentages of voc-ed students who transfer to a UC or CSU campus, earn a degree or certificate, or join the military; • Placement and retention of voc-ed students in UI-covered employment; and • Percentages of underrepresented gender students participating in voc-ed programs leading to nontraditional employment. It is also worth noting that the federal WIA and VTEA acts require that data be collected by program area. Hence, at the state level, VTEA and PFE data are available for California community colleges by program. This level of detail allows college administrators to conduct 55 internal program evaluation. Similarly, students have the information necessary to make an informed comparison of different colleges that offer the same programs and of different programs offered by the same college. Guidance to Policymakers for Evaluating Community Colleges Given the number and variety of performance standards they face, community colleges must shoulder a substantial burden connected with collecting and reporting required information. Added to this burden is the possibility of failing to meet one or more of the performance criteria and losing state and federal funding. As might be expected in these circumstances, community college administrators are vocal in their criticism of existing performance standards and not at all reticent in offering suggestions for how these standards might be revised. These criticisms include the following: • The costs of collecting required data are high. Especially burdensome are the requirements of following former students for up to six months and collecting and reporting data by program for each institution. • The standards are poorly designed to capture the range of successful student outcomes given the range of student backgrounds and interests. • Standards fail to reflect the multiple missions of community colleges. • Standards are likely to have unintended and perverse consequences. These include the dropping of worthwhile courses that do not figure into the reward system, failing to admit students who are less likely to graduate or to be easily placed in jobs or in four-year colleges, and grade inflation to boost completion and retention rates. If carried to an extreme, these criticisms suggest that each community college should be evaluated individually in terms of its performance in meeting the needs of the local community. A reasonable evaluation instrument for this purpose would be focus group interviews, 56 and we mentioned that a WIA performance indicator is consumer satisfaction with training services received. Nevertheless, the clear intent of legislators at both the federal and state levels is to mandate accountability standards requiring minimum levels of performance that are applied consistently across all colleges. It is a difficult task to establish performance standards that insure minimum levels of performance while recognizing that community colleges may have different missions. Our objective in this report is not to provide specific evaluation guidelines. Rather, our more modest objective is to develop evidence that informs policymakers, helping them to establish evaluation standards that contribute to improved performance without being viewed by community colleges as onerous or even counterproductive. What Does Our Evidence Show? We believe that the following observations drawn from our empirical results will be helpful to policymakers in formulating standards used in evaluating community colleges. 1. On average, California community colleges offer more credits in transferable programs than in other broad curriculum categories. In Chapter 2, we discussed PFE data on the distribution of credits offered, broken down by transferable, nontransferable voc-ed, adult basic skills, and other credits; and we noted that, on average across CCCS campuses, colleges offer 73 percent of total credits in transferable programs. Further analysis in Chapter 3 indicated that most colleges offer at least half of all credits in transferable programs. Moreover, the interquartile range for the proportion of transferable to all credits is found to be quite narrow, lying between 69 percent and 79 percent. With a few exceptions, our evidence suggests that California community colleges are heavily engaged in offering transferable curriculums. 2. Transferable curriculums are a mix of voc-ed courses and traditional academic courses. Simply comparing transfer rates, as in a recent study by Ehrenberg and Smith (2004), may be misleading. Transferable curriculums are typically thought to be composed of traditional academic courses, but as we pointed out in Chapter 2, they are in fact a mix of credits generated by transferable voc-ed courses as well as traditional 57 academic courses. Broadly defined voc-ed that includes both transferable and nontransferable voc-ed credits is, on average across CCCS campuses, about 23 percent of all credits. Of all voc-ed credits, in turn, about 64 percent are on average transferable credits. In Chapter 2, we focused on Santa Barbara City College as an example of a CCCS college that promotes itself as having a strong transfer orientation. It is interesting to note that at 77 percent, the ratio of transferable voc-ed to all voc-ed credits for SBCC is even higher than the all-college average (73%). Because skills acquired in transferable voc-ed programs are likely to be valued in local labor markets, a successful outcome for a college with a strong transferable voc-ed orientation might well be placement in a training-related job as opposed to a successful transfer to a four-year college. As indicated by Grubb (1996), probably the best way to measure the labor market effects of voc-ed curriculums is to use matched data that link student records to UI wage records. At the federal level, WIA and VTEA mandate the use of merged files containing student records and UI wage records to calculate employment, wage, and job retention outcomes. In their recent WIA implementation report, however, Macro, Almandsmith, and Hague (2003, chapter 8) indicate that few states have data currently available to evaluate the performance of training providers. Thus, performance standards specified in terms of labor market outcomes by program are likely to require the funding of additional institutional research by community colleges. Short of this financial commitment, the state might consider taking on centrally the management of large and complex administrative datasets and the production of the required data. 3. Although most credits offered by most colleges are transferable, there are important differences across CCCS campuses in curriculum “emphasis” and “specialization.” Information on differences in the missions of community colleges may be obtained by reading colleges’ promotional materials or by interviewing college officials in site visits. Our primary approach, in contrast, is to measure differences in the curriculum mixes offered to students. We argue that important differences in these mixes may be captured empirically by differences in curriculum “emphasis” and “specialization.” What we mean by curriculum emphasis is systematic 58 deviations from the mean of a particular curriculum measure. As reported in Chapter 3, the data suggest that colleges differ in the emphasis they place on voc-ed offerings, especially in the mix of voc-ed credits and courses that are transferable or taught at an advanced level. We also find that colleges differ in terms of the interest of their freshmen students in transferable versus voc-ed programs. Our definition of curriculum specialization goes beyond emphasis to link emphasis on one type of curriculum with a de-emphasis on another. Making use of this definition, we find that some colleges offer a transfer specialization including both traditional academic and voc-ed curriculums, whereas others specialize in nontransferable voc-ed. The number of colleges that specialize on these two dimensions depends on exactly how we implement our definition of specialization, and for this we consider two alternatives. Using our preferred alternative, we find that 26 colleges have a transfer curriculum specialty and another 19 colleges specialize in nontransferable voc-ed. Viewing these differences in curriculum emphases and specializations as indicating important differences in missions, we conclude that a “one-size-fits-all” evaluation strategy may not be appropriate. 4. Basic skills programs are a small proportion of most community colleges’ total offerings. Employers frequently cite lack of basic skills as a major obstacle to finding qualified employees, and adult basic skills curriculums are closely linked to community colleges’ workforce development mission. In Chapter 2, we reported that on average about 11 percent of full-time freshman students indicate that their primary academic objective is to acquire basic skills. Measured in terms of credits, about 7 percent on average of all credits are offered in basic skills courses. Although all colleges report freshman student interest in basic skills and offer basic skills courses for credit, we found in Chapter 3 that variation across colleges in our measures of basic skills is limited, especially in terms of credits offered. Our data thus indicate that the mission to provide basic skills courses is not one that receives great prominence, nor is it one on which community colleges attempt to differentiate themselves. Note that this conclusion rests on the assumption that the primary business of community colleges is to offer courses that supply credits toward a degree. 59 5. Colleges are responsive to differences in community characteristics in the curriculum mixes they offer. In Chapter 4, we examined the question of whether differences across colleges in curriculum emphases and specializations can be explained by indicators of college-specific characteristics and measures of community needs. College-specific characteristics include proximity to a four-year college and membership in a community college district, whereas community needs are measured by characteristics of the local labor market and community demographics. Our evidence indicates clearly that differences between colleges in curriculum mix are not randomly determined. In terms of collegespecific characteristics, we find that curriculum specializations and emphases have the expected relationships with membership in a multicampus district and proximity to a UC or CSU campus. Similarly, these measures of specializations and emphases appear to be related to the racial and ethnic composition of both students and the local community, in addition to other community demographic characteristics such as proportion of residents holding a bachelor’s degree. The conclusion we draw from this evidence is that California community colleges are responding in expected ways to differences in the interests and backgrounds of their students and to differences in the needs of their communities. In terms of accurately evaluating performance, these differences in missions should be taken into account. Allowing for Differences in Missions Precisely how to allow for differences in missions in measuring performance is a difficult issue that is currently receiving attention by policymakers in states across the nation. We conclude our report by commenting briefly on two possible approaches. The first was provided in the Job Training Partnership Act, which regulated federal employment and training activities for most of the 1980s and 1990s. The model-based JTPA procedure allowed performance standards to differ across colleges in response to quantitative differences in student characteristics and local economic conditions. Our evidence suggests that the model-based approach has merit. We find that differences in student characteristics and local economic conditions are closely related 60 to differences in colleges’ missions, as measured by differences in curriculum mix. Nevertheless, the model-based JTPA adjustment procedure was dropped from the Workforce Investment Act in 1998. In its place, WIA substituted a more ad hoc procedure that permits states either to waive performance standards or to negotiate adjustments in expected levels of performance. Such adjustments may be permitted if a college can make the argument that “unanticipated circumstances” involving a change in economic conditions or student characteristics make performance standards unreasonably burdensome. Analysts such as Barnow and Smith (2004) suggest that a fairer and less ad hoc procedure would be to return to the model-based system used under JTPA. A second approach involves up-front recognition by the state that community colleges differ in their missions. That is, rather than adjusting levels of performance expected for a given set of accountability standards, community colleges would be allowed to report data for those performance standards deemed to be consistent with their missions. There is some evidence that this approach may be receiving consideration at the federal level. First, the Employment and Training Administration (ETA) of the U.S. Department of Labor recently recommended reducing the 17 core WIA performance measures to a set of eight “common” measures, four of which would apply to adult programs and four to youth programs (Employment and Training Administration, 2003). Second, the U.S. House of Representatives, building on a Bush administration proposal, recently crafted and passed a WIA reauthorization bill (H.R. 1261). The reauthorization bill follows the ETA recommendation to reduce the number of performance criteria while giving governors authority to modify statewide standards in response to recommendations of local workforce investment boards and training service providers. Should California policymakers choose to take this approach to performance evaluation, we hope that the methodology developed in this report is helpful in measuring differences across community colleges in their missions. 61 Appendix A Technical Detail on Measures of Curriculum Mix PFE Data on Enrollment Partnership for Excellence data provide breakdowns of student credit hours attempted and completed by (1) type of course categorized by transferable, voc-ed, and basic skills, and (2) type of voc-ed program categorized as apprenticeship, advanced occupational, and introductory occupational. In the first or “Goal 3” breakdown, transferable courses are defined as courses that are transferable to UC or CSU campuses. Voc-ed courses, in contrast, are defined as nontransferable courses with Student Accountability Model (SAM) Priority Codes A, B, and C, classifying courses as “apprenticeship,” “advanced occupational,” and “clearly occupational,” respectively. The basic skills classification includes courses that supply precollege basic skills and basic skills unrelated to college entrance. In the PFE dataset, basic skills courses are assigned SAM Priority Codes D and E, indicating that the courses are “possibly occupational” and “nonvocational,” respectively. The Goal 3 breakdown of credits has the advantage of supplying a comprehensive overview of credits, but it provides a “narrow” specification of voc-ed. The second or “Goal 4” breakdown of voc-ed course enrollment retains the Goal 3 restriction limiting voc-ed courses to those with SAM Priority Codes A, B, and C. However, voc-ed courses are not restricted to nontransferable courses, thus allowing a “broad” specification of voced that includes transferable courses. The curriculum mix variables described in Table 2.2 include both narrow and broad measures of voc-ed. 63 VTEA Data Measuring Courses and Programs The federal VTEA and California state law require that each community college voc-ed program is to be evaluated using four core indicators of performance. For this purpose, the CCCS Chancellor’s Office maintains an electronic master Taxonomy of Programs (TOP) code file that lists over 300 individual programs and assigns to each a unique TOP code number. Vocational courses are defined as those having SAM Priority Codes A, B, C, or D and a vocational TOP code. Our data measuring voc-ed courses come from VTEA “Core Indicator Reports” using the “Course Voc Status/SAM Code” file. This file provides the number of voc-ed course offerings with SAM Codes A, B, C, and D for each community college in 1998–1999. In Table 2.3, the curriculum mix variables we report for voc-ed courses are constructed using data for SAM Codes A, B, and C. Total voc-ed courses are defined as the sum of SAM Codes A, B, and C courses. We use the same VTEA “Course Voc Status/SAM Code” file to measure the number of voc-ed programs offered grouped by SAM Priority Codes A, B, and C. Courses in each SAM Priority Code are listed by a TOP code. Our procedure for arriving at a number of programs in each priority code is to go down the list of courses offered and then to sum up the number of courses with different TOP codes. Two points should be noted regarding these data. First, since TOP codes are matched to courses, what we measure are program “areas” defined by content, which may not be the same as programs officially described in college catalogs. Second, our measures of SAM Codes A, B, and C programs are not mutually exclusive. The reason is that a program will often include courses categorized at more than one SAM Code level. For example, a program with a particular TOP code number might include advanced occupational (SAM Code B) courses as well as beginning occupational (SAM Code C) courses. Hence, this TOP-coded program would be counted in our dataset as both a SAM Code B program and a SAM Code C program. Chapter 2 notes that we do not have a reliable measure of total voc-ed programs with VTEA data. 64 Program Inventory Data In addition to VTEA data on voc-ed programs offered, the Chancellor’s Office maintains an “Inventory of Approved and Projected Programs.” The program inventory dataset includes all programs that are “approved” in the sense that a program must require 18 or more credits of course work. Hence, the number of programs included in this dataset is likely to understate the number of programs described in college catalogs, whereas, as noted, VTEA data report program areas measured by content rather than programs as they are officially titled. Each approved program is listed in the program inventory data by TOP code, and the Chancellor’s Office provides a list of programs matched to TOP codes that singles out vocational programs. Total approved voc-ed programs are also provided. First-Time Freshman Data As described in Chapter 2, FTF data are derived from a longitudinal study following the universe of first-time freshmen at all CCCS campuses between their initial enrollment in Fall 1997 and Spring 2000. Detailed information available by campus is presented in three categories: cohort, awards granted, and transfers. Cohort information includes gender, age, and race/ethnicity of students. In addition, FTF cohort data include a breakdown of freshman students’ academic objectives uninformed by counseling (“uninformed goals”) and informed by counseling (“informed goals”). Cohort data measuring students’ academic objectives uninformed by counseling are used in Chapter 2 to represent the demand of incoming students for alternative categories of educational services (voc-ed, transfer, basic skills, and other). FTF awards-granted data include total awards earned over the 1997–2000 period for the 1997 freshman cohort broken down by gender and race/ethnicity. Information is also available on type of award (A.A., A.S., and certificates) and title of program completed. Finally, FTF transfer data include total number of transfers broken down by whether the destination college is public or private, in California or out of state, the semester of transfer, and the name of the destination fouryear college or university. 65 Appendix B Technical Detail on Factor Analysis We implemented our factor analysis using as a guide the multistep strategy outlined by Johnson and Wichern (1988, p. 415). The software we used is the factor procedure in base SAS (SAS Institute, 1990). Johnson and Wichern recommend, as a first step, performing a principal component factor analysis with varimax (orthogonal) rotation. Our varimax rotation results yielded six common factors based on the criterion that the eigenvalues of the correlation matrix are greater than 1. Inspection of these results shows that the first four common factors extracted have a clear interpretation in terms of our curriculum mix variables. Moreover, each possesses the desirable feature of having at least three curriculum mix variables with large factor loadings. The remaining two common factors were both difficult to characterize and had fewer than three variables with substantial factor loadings. Since the last two common factors added little to our understanding, we restricted further estimations to four common factors. Factor loadings obtained using a varimax rotation for our 18 independent curriculum measures are reported in Table B.1. Continuing to follow the suggestions of Johnson and Wichern, we checked the sensitivity of our factor loadings to outliers, and we relaxed the orthogonality restriction by allowing for correlation among factors using the promax oblique (nonorthogonal) rotation. The main outlier we have identified is Santiago Canyon College with its heavy emphasis on apprenticeship training. Its omission caused only minor changes in the size of our factor loadings, so we retained Santiago Canyon in our analysis. Similarly, the promax rotation yielded results very similar to those obtained from the varimax rotation with little evidence of significant correlation between factors. 67 Table B.1 Factor Loadings from the Orthogonal Rotated Factor Patterns Curriculum Measure Transferable credits/ all credits Academic transfer credits/all credits Nontransferable voc-ed credits/all credits Nontransferable voc–ed credits/all voc-ed credits Interest in basic skills Advanced occupational credits/all voc-ed credits Advanced occupational courses/all voc-ed courses Advanced courses/ advanced programs Interest in voc-ed Interest in nontransfer programs Basic skills credits/ all credits Interest in transferring Apprenticeship courses/all voc-ed courses Apprenticeship credits/ all voc-ed credits Apprenticeship programs/100 students Voc-ed courses/100 students Advanced occupational programs/100 students Transferable Curriculum (1) .912 .801 –.906 –.711 –.677 .166 .289 –.019 –.053 .013 .041 .358 –.002 –.037 –.069 –.345 –.111 Advanced Voc-Ed Curriculum (2) .164 .050 –.105 –.176 –.044 .845 .841 .832 –.323 –.310 .069 .261 –.027 –.142 .168 .073 .195 Demand for Nontransferable Curriculum (3) –.060 –.230 –.149 –.128 .097 .138 –.083 –.108 .779 .777 .652 –.597 –.093 –.080 .114 .346 .413 Apprenticeship Training (4) –.096 .002 .149 .240 –.263 .022 –.031 .034 .230 .265 –.023 .051 .936 .862 .624 –.058 –.165 68 Table B.1 (continued) Curriculum Measure Approved voc-ed programs/total approved programs Variance explained (%) Transferable Curriculum (1) Advanced Voc-Ed Curriculum (2) –.018 24.5 –.029 13.2 Demand for Nontransferable Curriculum (3) –.073 12.7 Apprenticeship Training (4) .027 10.6 NOTES: Omitted curriculum variables are residual credits to all credits, undecided first-time freshmen to all freshmen, and broadly defined voc-ed credits to all credits; Copper Mountain College is omitted from the analysis. Factor loadings greater than 0.5 in absolute value are in bold. Finally, Johnson and Wichern suggest performing a maximum likelihood factor analysis including a varimax rotation. We were unable to successfully carry out the maximum likelihood analysis because the estimator gave communality estimates (squared multiple correlations of the common factors with the ith variable) that are greater than 1. This situation, described in the statistical literature as an “ultra-Heywood case,” is a particular problem when using the maximum likelihood method because variables with high communality are given high weights in subsequent iterations. The factor loadings shown in Table B.1 can be used to calculate predicted values of the common factors for each community college in our dataset. These predicted values are called factor scores. Descriptive statistics for the factor scores calculated for each of our four factors are shown in Table B.2. Note that factor scores are standardized to mean 0 and variance 1. Hence, the maximum/minimum and interquartile ranges shown in the table can be interpreted in terms of standard deviations. Colleges with the top five and bottom five factor scores for Factors 1–4 are listed in Table 3.5. 69 Table B.2 Descriptive Statistics for Factor Scores Obtained from Factor Analysis Common Factors Factor 1: transferable curriculum Factor 2: advanced vocational curriculum Factor 3: demand for nontransferable curriculum Factor 4: apprenticeship training Minimum Maximum –6.61 1.80 –1.36 4.50 –2.04 –1.37 3.04 7.77 Max/Min Range 8.46 IQ Range 0.89 5.86 1.23 5.07 1.23 9.14 0.57 NOTES: Factor scores are standardized to 0 mean and unit variance. Copper Mountain College is omitted from the analysis. 70 Appendix C Data Sources and Technical Details Used to Construct the Explanatory Variables College-Specific Characteristics Table 4.1 describes four college-specific variables. Beginning with multicollege district, PFE data on number of colleges in the district (Var9) are used to create a dummy variable indicating whether the college is part of a multicampus community college district. We used a three-step procedure to calculate proximity to the nearest UC or CSU campus. There are 10 campuses in the UC system and 23 campuses in the CSU system. Making use of a large map of California, we first identified at least two UC campuses and at least two CSU campuses as potentially nearest to each community college. For community colleges in large metropolitan areas such as Los Angeles and San Francisco, we picked up to seven UC campuses and seven CSU campuses for comparison. In the second step, we used Yahoo maps to calculate for each UC-CC pair and each CSU-CC pair (1) driving distance in miles and (2) travel time in minutes. The third step in the procedure involved choosing for each community college the nearest UC and the nearest CSU in driving distance and driving time. Proximity measured in miles is reported in Table 4.1. Campus age is obtained from the web site for each college. We strived to pin down the year the campus was founded, even if the college was originally a branch campus or subsequently changed its name. Student demographic data come directly from CPEC. 71 Local Job Opportunities Using online data from the 1997 Economic Census, we sought to obtain information about employers in the immediate proximity of the community college campus. Logically, this is best accomplished by selecting as the local labor market the city in which a community college is located. However, this characterization of the local labor market does not work for cities that are large enough to include multiple colleges. Nor does it work for communities that are too small to have the required Census data. In an attempt to deal with these problems, we used the following three-step protocol. 1. Check the city named in the college’s mailing address. If there is only one college in this city and Census data are available, use these data for this city. 2. If the city indicated in the mailing address is so large that it includes multiple colleges, check the five-digit zip code in the college’s mailing address. If the required data are available for this five-digit zip code, use these data. 3. If the city indicated in the mailing address is too small to have the required Census data, find the closest city for which data are available and use these data. Step 1 of the protocol is satisfied for the vast majority of colleges. We specified Step 2 to investigate further the case of colleges that, according to their mailing addresses, are in large cities served by multiple community colleges. Our hope was that by looking at a college’s fivedigit zip code, we could locate a smaller community within the larger metropolitan area for which required data are available. As it turned out, only for Los Angeles Southwest College were we successful in applying this step. For Los Angeles Southwest College, the city specified in the mailing address is Los Angeles, but the five-digit zip code indicates that the college is in Gardena. Since the required data are available for Gardena, we use employment data for Gardena rather than for Los Angeles. We were able to apply Step 3 to nine colleges whose local addresses indicate a small community in either a rural area or a large metropolitan 72 area. An example of the former is Cabrillo College in the small community of Aptos in Santa Cruz County. For this college, the closest larger city for which data are available is Santa Cruz. An example of the latter is Los Angeles Mission College in Sylmar. The required data are not available for Sylmar, but we could find data for nearby San Fernando. Overall, our protocol allows us to obtain unique employment data for 94 colleges. That is, we are forced to use duplicate data for 14 campuses. Los Angeles employment data are used for four Los Angeles area schools (out of nine Los Angeles Community College District campuses), and Oakland employment data are used for two Oakland area campuses (out of four campuses in the Peralta Community College District). Similarly, Sacramento employment data are used for all three colleges in the Los Rios Community College District serving Sacramento, San Diego employment data are used for all three colleges serving that city in the San Diego Community College District, and San Jose employment data are used for both colleges in the San Jose/ Evergreen Community College District. Local Service Area Demographics Our objective in specifying this set of variables is to measure for each college the demographic characteristics of its “local service area” defined in terms of prospective students, their parents, and local public officials who might be influential in determining curriculum mix. We considered several alternative approaches to defining the local service area. Initially, the approach that seemed the most promising was to make use of data measuring for each college the three-digit and five-digit origination zip code of all students enrolled in a community college as of Fall 2001. These data were kindly made available to us by Patrick Perry, Vice President for Management Information Systems in the Chancellor’s Office in Sacramento. Using 2000 Census data, the demographic information we require is also available at the three-digit and five-digit zip code levels, as well as for city and county. What we hoped was to find the zip codes (at the five-digit level or else at the three-digit level) that supplied a majority of a college’s students. Using this information to construct weights, we planned to calculate as a weighted average the 73 demographic characteristics of the local service area for each college. Unfortunately, we ran into the following problem. Enrollment at many colleges is so dispersed at the five-digit level that we were unable to isolate a reasonable number of five-digit zip codes that supply anywhere near a majority of student enrollment. On the other hand, the geographic areas covered by the relevant three-digit zip code were often so large that they failed to correspond to a reasonable definition of a local service area. Rather than trying to pin down the local service area in terms of three-digit and five-digit zip codes, the approach we ultimately followed was simply to make use of demographic data from the 2000 Census using information for the city in which the college is located. As indicated in Table 4.1, our community demographic variables include percentage of bachelor’s degree holders, percentage of foreign-born residents, median household income, and race and ethnicity. The city corresponding to each college, in turn, is specified using the protocol just outlined to measure local employer characteristics. 74 Appendix D Dataset for California Community Colleges Table D.1 Data Descriptions, Sources, and Descriptive Statistics Variable Var1 Var2 Var3 Var4 Var5 Var6 Var7 Var8 Var9 Var10 Var11 Var12 Var13 Var14 Var15 Var16 Var17 Var18 Var19 Var20 Description Attempted transfer, count enrollment Attempted basic skills, count enrollment Attempted nontransferable voc-ed, count enrollment Attempted all, count enrollments Attempted SAM Code A, count enrollment (apprenticeship) Attempted SAM Code B, count enrollment (advanced occupational) Attempted SAM Code C, count enrollment (clearly occupational) Attempted total voc-ed, count enrollment No. of colleges in district No. of SAM Code A courses No. of SAM Code B courses No. of SAM Code C courses No. of SAM Code D courses No. of SAM Code A programs No. of SAM Code B programs No. of SAM Code C programs No. of SAM Code D programs Total voc-ed programs Total “approved” voc-ed programs Total “approved” programs Data Source PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file PFE data file VTEA VTEA VTEA VTEA VTEA VTEA VTEA VTEA VTEA Program inventory Program inventory Mean 56009 Min. Max. 2720 156094 4851 287 16819 5737 0 25573 74525 317 10128 184948 0 12667 3578 0 25734 12029 1360 46076 15924 1520 48866 2 19 17 0 195 148 4 994 403 29 1877 137 0 695 2 0 20 23 1 71 35 13 82 24 0 80 41 15 96 34 10 64 50 16 99 75 Table D.1 (continued) Variable Var21 Var22 Var23 Var24 Var25 Var26 Var27 Var28 Var29 Var30 Var31 Var32 Var33 Var34 Var35 Var36 Var37 Var38 Var39 Var40 Var41 Var42 Var43 Var44 Var45 Var46 Var47 Var48 Var49 Var50 Var51 Var52 Var53 Var54 Var55 Var56 Var57 Var58 Var59 Description Total student headcount, Fall 2001 Formulate career interest Voc-ed degree, no transfer Acquire job skills Update job skills Voc-ed certificate Maintain license Degree and transfer Transfer, no degree Degree, no transfer Basic skills Educational development Complete GED Undecided Uncollected Total students surveyed Community college emphasis (voc-ed = 1, transfer = 2, indeterminate = 3) Year college founded No. of male students, Fall 2001 No. of female students, Fall 2001 No. of students gender unknown No. of Asian students No. of black students No. of Filipino students No. of Hispanic students No. of Native American students No. of white students No. of nonresident alien students No response on race/ethnicity Other race/ethnicity No. of full-time students No. of part-time students Full-time/part-time unknown Community college district code Miles to nearest UC campus Miles to nearest CSU campus Minutes to nearest UC campus Minutes to nearest CSU campus No. of employees in manufacturing Data Source www.cpec.ca.gov FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 FTF 1997 CPEC campus descriptions www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.cpec.ca.gov www.ccco.edu www.yahoo.com www.yahoo.com www.yahoo.com www.yahoo.com 1997 Economic Census Mean 14377 89 55 149 88 72 25 626 204 102 64 94 32 404 75 2077 3 1949 6333 7960 84 1592 1042 469 3756 143 5942 225 932 276 3295 8781 2402 37 48 27 58 35 15666 Min. 1993 0 0 0 0 0 0 35 6 2 0 0 0 0 0 182 1 Max. 38412 299 451 943 988 517 112 2111 1912 1030 533 847 250 1771 1022 5102 3 1883 881 1111 0 27 13 7 77 16 165 0 1 0 279 661 175 1 1.7 1.5 2 3 0 2001 17580 21109 773 9955 5417 2590 19518 646 22859 2953 5277 1513 8625 23179 9703 72 282.5 178.2 334 170 186758 76 Table D.1 (continued) Variable Var60 Var61 Var62 Var63 Var64 Var65 Var66 Var67 Var68 Var69 Description No. of employees in wholesale trade No. of employees in retail trade No. of employees in real estate No. of employees in professional services No. of employees in administrative and support services No. of employees in educational services No. of employees in health care No. of employees in arts and entertainment No. of employees in accommodations and food services No. of employees in other services (except public administration) Data Source 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census 1997 Economic Census Var70 Var71 Var72 Var73 Var74 Var75 Var76 Var77 Var78 Var79 Var80 Var81 Var82 % of population age 25+ with bachelor’s degree or higher % of population foreign born Median household income ($) Completion rate Transfer rate % white % black % Native American % Asian % Native Hawaiian and other Pacific Islander % some other race % two or more races % Hispanic ethnicity 2000 Census 2000 Census 2000 Census STRK STRK 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census 2000 Census Mean 7103 Min. Max. 10 87405 11377 10 118117 2775 10 39094 8978 10 122686 9479 25 104788 406 0 4599 6478 3 74565 1566 0 23403 9756 10 103746 2760 0 33761 26.5 5.4 78.1 23.3 49324 33.0 25.0 62.1 6.3 1.0 11.2 0.4 1.3 21900 11.5 0.0 16.7 0.2 0.1 0.3 0.1 54.4 173570 100.0 55.9 93.7 40.3 4.2 61.8 2.9 14.2 0.5 49.4 4.8 1.9 10.1 28.1 2.2 89.3 77 Table D.2 Construction of Curriculum Mix Variables from Raw Data Curriculum Variable Transfer credits/all credits Narrow voc-ed credits/all credits Basic skills credits/all credits Residual credits/all credits Broad voc-ed credits/all credits Narrow voc-ed credits/all voc-ed credits Apprenticeship credits/all voc-ed credits Advanced voc-ed credits/all voc-ed credits Academic transfer credits/all credits Apprenticeship courses/all courses Advanced voc-ed courses/all voc-ed courses Voc-ed courses/student Apprenticeship programs/student Advanced voc-ed programs/student Advanced courses/advanced programs Approved voc-ed programs/total programs Undecided student interest/all students Student interest in basic skills/all students Student interest in transferring/all students Student interest in voc-ed/all students Student interest in nontransfer programs/all students Transformation of Raw Data Var1/Var4 Var3/Var4 Var2/Var4 (Var4–Var1–Var2–Var3)/Var4 Var8/Var4 Var3/Var8 Var5/Var8 Var6/Var8 [Var1–(Var8-Var3)]/Var4 Var10/(Var10+Var11+Var12) Var11/(Var10+Var11+Var12) (Var10+Var11+Var12)/Var21 Var14/Var21 Var15/Var21 Var11/Var15 Var19/Var20 Var34/(Var36-Var35) (Var31+Var32+Var33)/(Var36-Var35) (Var28+Var29)/(Var36-Var35) (Var22+. . .+Var27)/(Var36-Var35) (Var22+. . .+Var27+Var30)/(Var36-Var35) 78 References Bailey, Thomas, “Community Colleges in the 21st Century: Challenges and Opportunities,” in Patricia A. Graham and Nevzer G. Stacey, eds., The Knowledge Economy and Postsecondary Education: Report of a Workshop, National Academy Press, Washington, D.C., 2002, pp. 59–75. Barnow, Burt S., and Jeffrey A. Smith, “Performance Management of U.S. Job Training Programs,” in Christopher J. O’Leary, Robert A. Straits, and Stephen A. Wandner, eds., Job Training Policy in the United States, W. E. Upjohn Institute for Employment Research, Kalamazoo, Michigan, 2004. California State Department of Education, Master Plan for Higher Education in California, 1960–1975, Sacramento, Calfornia, 1960. Chancellor’s Office, “Student Right-to-Know: Rates and Analysis for the 1999 Cohort,” Technology, Research, and Information Systems Division, California Community Colleges, Sacramento, California, June 20, 2003. Ehrenberg, Ronald G., and Christopher L. Smith, “Within State Transitions from 2-Year to 4-Year Public Institutions,” Economics of Education Review, Vol. 23, February 2004, pp. 11–28. Employment and Training Administration, “Common Measures Policy,” Training and Employment Guidance Letter No. 15-03, Washington, D.C., December 10, 2003. Grubb, W. Norton, Working in the Middle: Strengthening Education and Training for the Mid-Skilled Labor Force, Jossey-Bass, San Francisco, California, 1996. Gumport, Patricia H., “The Demand-Response Scenario: Perspectives of Community College Presidents,” Annals of the American Academy of Political and Social Science, Vol. 58, March 2003, pp. 38–57. 79 Johnson, Richard A., and Dean W. Wichern, Applied Multivariate Statistical Analysis, Prentice Hall, Englewood Cliffs, New Jersey, 1988. Joint Committee to Develop a Master Plan for Education, “The California Master Plan for Education,” Sacramento, California 2003. Kane, Thomas J., and Cecilia E. Rouse, “The Community College: Educating Students at the Margin Between College and Work,” Journal of Economic Perspectives, Vol. 13, Winter 1999, pp. 63–84. Laden, Berta V., “Two-Year Hispanic-Serving Colleges,” in Barbara Townsend, ed., Two-Year Colleges for Women and Minorities: Enabling Access to the Baccalaureate, Falmer Press, New York, 1999, pp. 151–194. Macro, Bronwen, Sherry Almandsmith, and Megan Hague, “Creating Partnerships for Workforce Investment: How Services Are Provided Under WIA,” revised final report, Berkeley Policy Associates, Oakland, California, September 2003. Romano, Richard M., “Financing Community Colleges Across the States: An Economic Perspective,” paper presented at the Cornell Higher Education Research Institute conference on “The Complex Community College,” Cornell University, Ithaca, New York, October 13–14, 2003. SAS Institute, SAS User’s Guide: Statistics, Version 5 edition, SAS Institute, Inc., Cary, North Carolina, 1990. State of California, “1996 Amendment to the California State Education Code: Chapter 1057,” Section 2, 1996. 80 About the Authors ANDREW M. GILL Andrew M. Gill is a professor of economics at California State University, Fullerton, where he specializes in labor economics and econometrics. His research explores community college issues, gender and racial discrimination, occupational wage differentials, and the labor market effects of drug use. He is currently co-editor of the journal Contemporary Economic Policy. He received his Ph.D. in 1985 from Washington State University. DUANE E. LEIGH Duane E. Leigh is a professor of economics at Washington State University, Pullman. Since the mid-1980s, his research has focused on the labor market effects of adult training and retraining programs designed for displaced workers and female welfare recipients. He received his Ph.D. in 1969 from Michigan State University. 81 Related PPIC Publications Financing California’s Community Colleges Patrick J. Murphy The Effects of School-to-Career Programs on Postsecondary Enrollment and Employment David Neumark School Finance and California’s Master Plan for Education Jon Sonstelie and Peter Richardson (editors), Julian R. Betts, Eric J. Brunner, Anne Danenberg, Jane Leber Herr, Susanna Loeb, Lawrence O. Picus, Peter Richardson, Heather Rose, Kim S. Rueben, and Jon Sonstelie (contributors) PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 83" ["post_date_gmt"]=> string(19) "2017-05-20 09:36:12" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_804agr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:36:12" ["post_modified_gmt"]=> string(19) "2017-05-20 09:36:12" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_804AGR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }